Forem: Nitya Narasimhan, Ph.D

Model Mondays S2E04 - AI Developer Experiences

Nitya Narasimhan, Ph.D — Fri, 11 Jul 2025 03:25:13 +0000

🧪 | Some parts of this blog were generated with AI assistance, and reviewed manually before publishing. To learn more about the why and the how, please refer to this document in our website.

Model Mondays is a weekly series that helps you build your AI Model IQ with 5-min news recaps, 15-min tech spotlights and 30-min AMA sessions with subject matter experts. Join us!

Register for upcoming livestreams (Mon @1:30pm ET)

Register for upcoming AMAs (Fri @1:30pm ET)

Subscribe to the weekly newsletter

Explore Season 1 & Season 2 episodes on our repo!

⁉️ | Have questions on the AI TOOLKIT EXTENSION FOR VS CODE

Download the AI Toolkit and try it!

Join the AMA on Friday, Jul 11 on the Azure AI Foundry Discord.

Visit recaps & resources on the Forum to keep learning.

Spotlight On: AI Developer Experiences

In this episode we got to see a hands-on demo from Leo Yao, a Product Manager for the AI Toolkit, who took us through the various tools and capabilities that this extension brings to Visual Studio Code, to streamline our AI development journey. If you missed the livestream, you can catch up on the recap here:

What You'll Learn

So why should you watch this episode? Simple! If you wanted to see how you could streamline your AI Developer Experience without ever leaving Visual Studio Code, then the AI Toolkit is the extension you want to explore. In the space of just 15 minutes, I got to see Leo complete the following tasks, all within the IDE!

Discover & use GitHub Models with a built-in playground feature.
Deploy models (GPT-4) to Azure AI Foundry directly from VS Code.
Build AI agents with Agent Builder (to createa a math tutor)
Configure agents with prompts & tools (for an escape room helper)
Integrate MCP servers to extend functionality (custom Python code)
Deploy agents locally, and to Azure AI Foundry (for flexibility)

And that was just in 15-minutes. I can only imagine what other features and tools we are yet to discover if we try this out ourselves. Want to get links to the extension, documentation and key resources to skill up? You'll find the links in the deck below:

How to Get Started

Looking back, my first introduction to the AI Toolkit was in Model Mondays Season 1, from Microsoft Product PM Rong Lu. She gave us a comprehensive overview of the various features including support for data generation and evaluation. I recommend taking 15 minutes to review her demo in this episode:

Once you're familiar with the basics, I recommend you check out these three links to dive deeper into the process of building agent-based solutions on Azure AI Foundry with AI Toolkit and Visual Studio Code.

1. AI Toolkit for Visual Studio Code - Getting Started Guide - This is my recommended starting point. The tutorial covers installation, then walks you through the steps for downloading models from the catalog, then testing and integrating models into your application. You can then keep going with more tutorials to learn other features like fine-tuning support.

2. Azure AI Foundry Extension for Visual Studio Code - This extension now lets you work with the Azure AI Foundry platform capabilities directly from VS Code. This screenshot from the docs gives you a sense of the tasks you can achieve from your IDE - including project setup and model deployments.

3. Agent Builder in AI Toolkit -The Agent Builder in AI Toolkit simplifies and streamlines your workflow for building agents - from prompt engineering to tool integration (including with MCP servers). Get an intuitive sense for going from model to prompt to agentic workflows with tooling help.

⁉️ | Have questions? Don't forget to join us for the AMA on Friday (Jul 11) on the Azure AI Foundry Discord. Post your questions early on the Forum or revisit it later to get the transcript and resources from the discussion!

Read The Blog

Want an easy way to catch up on all the news? Check out complete blog series written by our resident student blogger here on Tech Community!

You can also follow her posts, right here on dev.to:

Sharda KaurFollow

Gold @Microsoft MLSA || Technical Content Writer || Open Source Contributor || Web Developer || CUIET MCA'26

Build Your Model IQ!

Model Mondays Season 2 is currently scheduled to go from June to September, covering the 12 key topics shown below.

👉🏽👉🏽 Register for Upcoming Livestreams to get reminders
👉🏽👉🏽 Register for Upcoming AMAs to get reminders

Model Mondays S2E03 - SLMs and Reasoning

Nitya Narasimhan, Ph.D — Thu, 03 Jul 2025 03:01:19 +0000

Model Mondays is a weekly series that helps you build your AI Model IQ with 5-min news recaps, 15-min tech spotlights and 30-min AMA sessions with subject matter experts. Join us!

Register for upcoming livestreams (Mon @1:30pm ET)

Register for upcoming AMAs (Fri @1:30pm ET)

Subscribe to the weekly newsletter

Explore Seasom 1 & Season 2 episodes on our repo!

Spotlight On: SLMs and Reasoning

🧪 | This section was generated with AI help and human revision & review. To learn more, please refer to this document in our website.

How can you bring advanced reasoning to resource-constrained devices? This session explores the latest in Small Language Models (SLMs) like Phi-4, which are redefining what’s possible for agentic apps.

Mojan Javaheripi, a Senior Microsoft Researcher, will discuss how SLMs leverage inference-time scaling and chain-of-thought reasoning to deliver powerful results on smaller hardware. Discover use cases, deployment strategies, and how SLMs are making AI more accessible and efficient for everyone.

Download The Deck With Key Resource Links:

Phi-4 Reasoning: Power Through Precision

In her spotlight talk, Mojan introduced the Phi-4-reasoning model and a Phi-4-reasoning-plus model and walked us through the techniques used to create these variants in a manner that allows them to work in resource-constrained environments, while providing comparable accuracy to larger language models. The motivation for this work comes from wanting to bridge the gap between large frontier models and smaller, more efficient models that can run on commodity hardware including personal devices like laptops, but also edge devices at industry scale.

Watch The Replay From Her Livestream

Why is this important?

The Phi-4 reasoning models are designed for explicit, step-by-step problem solving, with a focus on explainability and logical decomposition. Mojan details the training pipeline, including data curation, supervised fine-tuning, and reinforcement learning, and shares benchmarking results that show Phi-4 models outperforming much larger models in math, science, and coding tasks. The discussion covers community adoption, quantization for edge devices, and the importance of fine-tuning for domain-specific applications. Real-world use cases include intelligent tutoring, autonomous agents, code generation, and strategic planning.

Key Takeaways From The Session

Phi-4 reasoning models deliver state-of-the-art reasoning and efficiency at a fraction of the size of frontier models.
SLMs are ideal for edge, local, and resource-constrained deployments, with strong support for quantization and customization.
Fine-tuning & reinforcement learning enable adaptation to new domains & tasks.
Open-source and community contributions accelerate innovation and adoption.
Responsible AI, safety, and explainability are core to Phi-4 model design.

Sidebar: AI & Content Generation

You may have noticed the 🧪 annotation about AI-generated content. Let's talk about how this was done, and why.

Model Mondays is a series about building your AI model IQ - and learning how to find the right model for the task. And the best way to learn something is by applying it to a real problem. Starting this week, I'm using AI to streamline many content-related tasks for Model Mondays. And I wanted to share not just the reasoning for this, but the process and tooling I use - in the hopes that it helps others find productive uses for the same workflows.

Motivation: Efficiency

On a weekly basis I have to plan for two different sessions (livestream and AMA), coordinate across speakers and hosts, then publish pre- and post- event content writeups that provide promotion and recap value respectively.

Writing these manually is not just time-consuming, it is also complicated by the fact that each week has new topics to research - and new discussions to summarize that require cognitive bandwidth. My a-ha moment was in recognizing that prompt creation is also content creation and crafting a well-written prompt to create content can be as rewarding and effective as authoring the content from scratch.

Workflow: Agent Mode + MCP

How am I doing this? I'm using Visual Studio Code Agent Mode in a GitHub Codespaces environment launched on the Model Mondays repo. Every task is reduced to a custom prompt that is stored in the codebase for reuse.

Here are some examples:

Give it the YouTube transcript and have it summarize it
Give it the AMA transcript and have it generate a recap
Detect key terms and use MCP Servers to find resource links.

You can simplify this further if you use MCP Servers to extend Agent Mode as shown - adding new "tools" to help the agent perform key tasks.

In my case, there were 3 tools that were immediately helpful:

Microsoft Docs MCP Server - for related resources
Hugging Face MCP Server - for Model Hub links
Azure AI Foundry MCP Server - for Azure AI Foundry Models catalog

Want to see these in action? Check out the "Related Resources" in this post, populated by using the three MCP servers above.

Keep reading this series for more updates and learnings from our AI experiments!

Read The Recap

Want an easy way to catch up on all the news? Check out complete blog series written by our resident student blogger here on Tech Community!

You can also catch her recap for this week, right here on dev.to:

Sharda Kaur

Jul 5 '25

I Wrote About SLMs in Model Mondays: Easy Explanation Inside!

#microsoft #beginners #ai

Comments

1 min read

Build Your Model IQ!

Model Mondays Season 2 is currently scheduled to go from June to September, covering the 12 key topics shown below.

👉🏽👉🏽 Register for Upcoming Livestreams to get reminders
👉🏽👉🏽 Register for Upcoming AMAs to get reminders

Model Mondays S2E02 - AMA on Model Context Protocol

Nitya Narasimhan, Ph.D — Tue, 24 Jun 2025 17:50:47 +0000

Model Mondays is a weekly series that helps you come up to speed with the fast-moving world of AI models. Here are 3 actions you can take to build your model IQ:

Watch Model Mondays Live - for news roundup & tech spotlights in 30 mins
Join Foundry Friday AMA - for discussions with Q&A featuring AI experts
Subscribe Model Mondays Newsletter - a weekly pulse on AI innovation & tech

Spotlight On: Model Context Protocol

We put the spotlight on Model Context Protocol (MCP) with a deep-dive from Den Delimarsky, a member of the MCP Steering Committee.

👉🏽👉🏽 Register for the AMA on our Azure AI Foundry Discord
👉🏽👉🏽 Post Your Questions on our Discussion Forum

Then check out the slides from the presentation, for resource links!

About Model Context Protocol (MCP)

MCP is a protocol that standardizes how AI applications connect the underlying AI models to required knowledge sources (data) and interaction APIs (functions) for more effective task execution. Because these models are pre-trained, they lack access to real-time or proprietary data sources (for knowledge) and real-world environments (for interaction). MCP allows them to "discover and use" relevant knowledge and action tools to add relevant context to the model for task execution.

Explore: The MCP Specification
Learn: MCP For Beginners
Watch: Anthropic Workshop on MCP (below), AI Engineer Summit NY 2025

Securing MCP

As MCP started gaining traction, the conversation turned to the topic of secure MCP usage, particularly in enterprise environments. How should MCP clients authenticate with MCP servers, and get authorization for access to the right resources? In early 2025, the conversation began on an MCP Authorization spec - here are the three documents that help you understand the issues & proposal process.

Aaron Parecki, Let's Fix OAuth in MCP
Den Delimarsky, Improving The MCP Authorization Spec - One RFC At A Time
MCP Specification, Authorization protocol draft

On Monday, Den joined us live to talk about the work he did for the authorization protocol. Watch the session now to get a sense for what the MCP Authorization protocol does, how it works, and why it matters. Have questions? Submit them to the forum or Jon the Foundry Friday AMA on Jun 27 at 1:30pm ET.

Learning MCP

Want to get hands-on experience with MCP from fundamentals to best practices? Explore the "MCP For Beginners" curriculum - a 10-lesson open-source curriculum that is constantly being updated to reflect the latest progress and examples of MCP in action. Go from fundamental concepts to best practices, with hands-on exercises.

👉🏽👉🏽 Explore the Curriculum on GitHub

Read The Recap

Want an easy way to catch up on all the news? Check out complete blog series written by our resident student blogger here on Tech Community!

You can also catch her recap for this week, right here on dev.to:

Sharda Kaur

Jun 29 '25

🚀 Model Mondays S2E2: Understanding Model Context Protocol (MCP) – A Beginner-Friendly Guide

#opensource #mcp #microsoft #beginners

Comments

1 min read

Continue Building Model IQ

Check out the upcoming episodes and register for livestreams & AMA:

👉🏽👉🏽 Register for Livestreams to get reminders
👉🏽👉🏽 Register for AMAs to get reminders

Model Mondays S2E01 - AMA on Advanced Reasoning

Nitya Narasimhan, Ph.D — Tue, 17 Jun 2025 23:44:52 +0000

Interested in regular Model News & Updates? Subscribe to the The Model Mondays Newsletter - read the first edition for context.

Model Mondays is a weekly series that combines a 30-minute livestream (Monday at 1:30pm ET) with a 30-minute AMA (Fridays at 1:30pm ET) to build your AI model IQ.

5-min Highlights - recaps the top 5 news items from last week
15-min Spotlights - a deep-dive with a subject matter expert
30-min AMAs - open-ended discussion, Q&A with the expert

This week we put the spotlight on Advanced Reasoning with subject matter expert Marlene Mhangami - and covered the top 5 model-related news from Azure AI Foundry - in our #ModelMondays livestream. Browse the slides below - then join us Friday for Q&A with Marlene:

👉🏽👉🏽 Register for the AMA on our Azure AI Foundry Discord
👉🏽👉🏽 Post Your Questions on our Discussion Forum

Want to get a sense for what Reasoning models are, what advanced reasoning covers? Read on.

What are Reasoning models?

Reasoning models are a new category of Large Language Models that are trained (with large-scale reinforcement learning) to think deeply before they respond. These reasoning abilities are achieved by a combination of techniques including chain-of-thought, self-consistency, and deliberative alignment.

Reasoning models are designed to tackle hard problems involving logic, stratgegy, complex reasoning and multi-step planning. They are particularly effective in STEM fields (math, science, coding) where they can outperform popular general-purpose models in competitive benchmarks.

Want an introduction to reasoning models? Watch this episode from Season 1 where Jennifer Marsman covers the OpenAI o-series of reasoning models.

Advanced Reasoning Scenario

While the previous talk gives you a sense for what reasoning models are with a real-world use case, this advanced reasoning episode takes it a step further with an end-to-end sample and workshop that teaches you how to build a Deep Researcher - a research assistant that can conduct comprehensive web research, analyze & synthesize the information, and present its findings in a report.

The solution uses the DeepSeek-R1 reasoning model with the Tabily web search API, and the LangGraph framework to perform iterative cycles of research before delivering the final result. Watch the livestream of the episode below:

Reasoning Lab For Beginners

Want to get hands-on experience with reasoning models and learn how they differ from general-purpose models? Try this self-guided lab from Microsoft Build 2025 which explores the OpenAI o-series of reasoning models with hands-on exercises!

👉🏽👉🏽 Explore The Lab on GitHub
👉🏽👉🏽 Browse The Presentation on SpeakerDeck

The storyboard summarizes what the lab covers.

Build Your Model IQ!

Model Mondays Season 2 is currently scheduled to go from June to September, covering the 12 key topics shown below.

👉🏽👉🏽 Register for Upcoming Livestreams to get reminders
👉🏽👉🏽 Register for Upcoming AMAs to get reminders

This Week In Model Mondays

Want an easy way to catch up on all the news? Check out complete blog series written by our resident student blogger here on Tech Community!

You can also catch her recap for this week, right here on dev.to:

Sharda Kaur

Jun 25 '25

I Wrote the S2E01 Recap for Model Mondays: Advanced Reasoning Session

#ai #microsoft #azure #beginners

Comments

1 min read

Continue Building Model IQ

Check out the upcoming episodes and register for livestreams & AMA:

👉🏽👉🏽 Register for Livestreams to get reminders
👉🏽👉🏽 Register for AMAs to get reminders

S1E1 - Hands-on With GitHub Models!

Nitya Narasimhan, Ph.D — Mon, 17 Mar 2025 02:08:25 +0000

Model Mondays is a weekly livestream series where we put the spotlight on one model or model theme each week, to help you build your model IQ! Plus, get model news highlights in our 5-min roundup.

Three Links To know:
1️⃣ | Watch The Livestream Mondays @ 1:30pm ET · RSVP
2️⃣ | Join The Chat Fridays @ 1:30pm ET · Discord
3️⃣ | Explore Resources for Hands-on Learning · GitHub

S1:E2: RSVP To The Next Episode

🌟 | Register Now and join us live on Monday, Mar 17 at 1:30pm ET

In this episode, we’re joined by Jennifer Marsman as we dive into the world of reasoning models. We’ll get hands-on with the o1 series of models and more as we learn what these models can do, how they work, and how you can use them most effectively for your generative AI application needs.

S1:E1: Recap Of The Last Episode

In our season kickoff episode, we put the spotlight on the GitHub Models Marketplace and showed you how you can kick off your hands-on learning journey into generative AI and language models with just a GitHub account.

Why use GitHub Models? - Simply put, they are an opportunity for you to explore and experiment with cutting-edge AI models, for free! Follow these instructions to get started today!

What can you do with GitHub Models? - The GitHub Models Marketplace provides a subset of the models on the Azure AI Foundry model catalog - for free - as serverless endpoints that you can invoke with just a GitHub Personal Access Token. You can then do the following:

Select a model and use the Playground to explore it interactively
Use the prompt editor to iterate on prototypes
Compare two models side-by-side to help with model selection
Get ready-to-use code snippets to access them from your IDE
Access them using the Visual Studio Code AI Toolkit extension
Save and share your experiments with others.

Watch the livestream recording to see quick demos of many of these features! Then visit the Model Mondays repo each week for updates. We'll have notebooks and additional resources for you to explore shortly.

Missed the livestream? Visit the event page to catch up on replay and resources.

Download The Presentation

Want to share the resources or do a talk about this to your community? Feel free to reuse the slides or share your own!

You can find the slides from this episode at the link below. Use the QR codes to explore the relevant models in the model catalogs, then dive into the GitHub Models spotlight segment to try out the exercises for yourself!

Follow-up Fridays: Community Chat

Have questions or comments? Have you built something with GitHub Models or want to share your insights working with a specific model family or provider? Join us on the Azure AI Discord every Friday at 1:30pm ET for a community chat - using this link

#ModelMondays - A Weekly Series To Build Our AI Model IQ

Nitya Narasimhan, Ph.D — Mon, 10 Mar 2025 03:00:07 +0000

Three Links To know:
1️⃣ | Watch The Livestream Mondays @ 1:30pm ET · RSVP
2️⃣ | Join The Chat Fridays @ 1:30pm ET · Discord
3️⃣ | Explore Resources for Hands-on Learning · GitHub

Why Model Mondays?

Have you felt overwhelmed by the non-stop drumbeat of AI model announcements? Are you feeling a sense of decision fatigue with making a simple model choice? You aren't alone!!

We all have similar questions here:

How can we catch up on updates without getting overloaded?
How can we pick the right model given thousands of choices?
How can we get hands-on experience with models to skill up?
How can we learn the tools & techniques for AI customization? This was the main motivation for kickstarting an 8-week series that we are calling Model Mondays!! And it launches tomorrow!!

What is Model Mondays?

Think of it as a weekly "power-up" jouney that takes you from awareness to actionable learning. Build your generative AI model IQ in 3 ways:

(Watch) 30-minute livestream every Monday @1:30pm EST · RSVP HERE
(Discuss) A 30-minute community chat every Friday @1:30pm EST · Join Discord
(Explore) A repository of resources for hands-on learning · Fork The Repo

Check out this fun 2-minute trailer I made, to give you a sense of what we will cover - then read on for more details on how you can get the most value from this initiative.

Watch and Learn: Model Mondays Livestream

We will be kicking off the series tomorrow so do join us! Each 30-minute episode is structured into 2 main segments, with time for discussion and Q&A at the end.

5-min Model Roundup

The 5-minute roundup is meant to grow information awareness without overload. We'll tell you about 5 model news items from the past week, and give you some context for why they matter. And, you can check out the episode pages on GitHub to explore added links from the community that you can read at a more leisurely pace!

15-min Model Spotlight

The 15-minute spotlight is meant to drive actionable learning about these new model releases and model categories - with hands-on demos and code samples or tutorials that you can try out yourself! In this first "season" we hope to cover:

GitHub + Models - for jumpstarting your exploration journey!
Reasoning + Models - for tackling complex problem-solving
Search & Retrieval + Models - for building better RAG solutions
Visual Generative + Models - for text-to-image and image-to-image
Fine Tuning + Models - for AI customization with efficient tooling
Synthetic Dataset + Models - for use in evaluations etc.
Open Source + Models - from open-weights to open-source software.
Forecasting + Models - for domain-specific tasks on time-series data.

Connect and Share: Discord Community

The best way to learn anything is to be in watercooler conversations with others who share your interests and bring their own lived experiences and insights. And this is where we hope you join us and participate in building community IQ around generative AI models.

Join the Azure AI Discord - Invite here
Use the #model-mondays channel to exchange ideas & ask questions
Show-and-Tell Community Chats - Fridays @1:30pm ET

Launching Mar 10! 🆕

We hope you find this series informative but also actionable. Don't forget to register below and join us for the kickoff tomorrow!

Three Links To know:
1️⃣ | Watch The Livestream Mondays @ 1:30pm ET · RSVP
2️⃣ | Join The Chat Fridays @ 1:30pm ET · Discord
3️⃣ | Explore Resources for Hands-on Learning · GitHub

New Azure AI tools to help operationalize Responsible AI for generative AI apps!

Nitya Narasimhan, Ph.D — Mon, 01 Apr 2024 05:45:07 +0000

Welcome to the 12th post in my This Week In AI News series. Each week I post 1 article on something I'm working on, or find valuable, for AI developers. Want an easier way to get new posts? Follow this tag: 👇🏽

#thisweekinai Follow

Key Resources To Know This Week

Resource	Description
1️⃣ Blog Post	Official Announcement of New Tools
2️⃣ Collection	My Responible AI For Developers Collection
3️⃣ Collection	My Azure AI For Developers Collection

🆕 | Watch the 2-part AI Show Episode on this topic!

Click each thumbnail image to view the full replay video.

We've talked about #ResponsibleAI in two contexts before - Model Debugging for predictive AI apps (MLOps), and AI-Assisted Evaluation for generative AI (LLMOps). Then late last week, I shared this exciting announcement about new Responsible AI tools coming to Azure AI!

// Detect dark theme var iframe = document.getElementById('tweet-1773403831050645539-630'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1773403831050645539&theme=dark" }

In today's post, I want to dig a little deeper into the announcement to learn what the tools do, why they matter, and how developers can get started using them in their generative AI application workflows.

1 | The Azure AI Platform

The Azure AI Studio provides a browser-based UI/UX for exploring the rich capabilities of the Azure AI plaform as shown below. It also supports a code-first experience for developers who prefer working with an SDK or command-line tools.

It streamlines your end-to-end development workflow for generative AI applications - from exploring the model catalog, to building & evaluating your AI application, to deploying and monitoring the application in production. With built-in support for operationalizing Responsible AI, developers can go from evaluating their applications for quality, to configuring them for content safety in production.

2 | New Azure AI Tools for Responsible AI

The recent announcement from the Responsible AI team highlights a number of new tools and capabilities that are available (or coming soon) to Azure AI, to further improve the quality and safety of generative AI application on Azure. This short video gives you a quick preview of how these tools are put to use to create safeguards for generative AI apps on Azure. In the rest of this post, we'll dive briefly into each of these tools to understand what they do, and why it matters.

Watch the 2-part series on The AI Show (links at the top of this post) for more details.

3 | Prompt Shields

The first new capability comes in the form of Prompt Shields that can detect and block prompt injection attacks to safeguard the integrity of your LLM system. These attacks work by tricking the system into harmful or unplanned behaviors, "injecting" unauthorized instructions into the default user prompt at runtime.

In a direct attack (jailbreak) the user is the adversary. The user prompt attempts to get the model to disregard developer-authored system prompts and training in favor of executing potentially harmful instructions. In an indirect attack (cross-domain prompt injection) the adversary is a third-party and the attack occurs via untrusted external data sources that may be embedded in the user prompt, but not authored by user or developer.

Prompt Shields work proactively to detect suspicious inputs in real-time and block them before they reach the LLM. This can use techniques like Spotlighting that transform the input to mitigate these attacks while preserving the semantic content of the user prompt.

🔖 | Learn more in this post
🔖 | Review the main announcement

4 | Groundedness Detection

The second capability involves Groundedness Detection to combat the familiar problem of "Hallucinations". Here, models fabricate a response that may look valid but is not grounded in any real data. Identifying and remediating this is critical to improve trustworthiness of generative AI responses.

Previously developer options included manual checks (not scalable) and chaining requests (to have an LLM evaluate if the previous response was grounded with respect to a reference document) with mixed results. The new tool uses a custom-built fine-tuned language model that detects ungrounded claims more accurately - giving developers multiple options to mitigate the behavior, from pre-deployment testing to post-deployment rewriting of responses.

🔖 | Learn more in this post
🔖 | Review the main announcement

5 | Safety System Messages

The third capability recognizes that prompt engineering is a powerful way to improve the reliability of the generative AI application, along with services like Azure AI Content Safety. Writing effective system prompts (metaprompts) can have a non-trivial impact on the quality of responses - and system messages that can "guide the optimal use of grounding data and overall behavior" are ideal.

With the new system message framework and template recommendations for LLMs, developers now get example templates and recommendations to help them craft more effective messages. For instance, the system message framework describes four concepts (define model capabilities, define model output format, provide examples, provide behavioral guardrails) you can apply in crafting the system message. The screenshot above shows an example of how this is applied in a retail chatbot app.

🔖 | Learn more from the documentation
🔖 | Review the main announcement

6 | Automated Safety Evaluations

The fourth capability recognizes that most developers lack the resources and expertise to conduct rigorous safety evaluations on their generative AI applications - which would involve curating high-quality datasets for testing, and interpreting evaluation results for effective mitigation.

Previously, Azure AI supported pre-built generation quality metrics like groundedness, relevance, coherence and fluency for AI-assisted evaluations. With the new capability, this now includes additional risk and safety metrics like hateful and unfair content, sexual content, violent content, self-harm-related content, and jailbreaks.

To conduct a safety evaluation on your generative AI application, you need a test dataset and some way to simulate adversarial interactions with your application so you can evaluate the resulting responses for the relevant safety metrics. The new Azure AI automated safety evaluations capability streamlines this for you in four steps:

Start with targeted prompts (created from templates)
Use AI-assisted simulation (for adversarial interactions)
Create your test datasets (baseline & adversarial)
Evaluate the test datasets (for your application)

Outcomes can now be used to configure or adapt other elements of the application's risk mitigation system.

🔖 | Learn more in this post
🔖 | Review the main announcement

7 | Risk & Safety Monitoring

The final new capability announced was around Risk & Safety Monitoring in Azure Open AI - adding a new Dashboard capability described as follows:

In addition to the detection/ mitigation on harmful content in near-real time, the risks & safety monitoring help get a better view of how the content filter mitigation works on real customer traffic and provide insights on potentially abusive end-users.

To use the feature, you need an Azure OpenAI resource in a supported region, and a model deployment with a content filter configured. Once setup, simply open the Deployments tab, visit the model deployment page, and select the Risks & Safety tab as shown in this figure from the announcement post below.

The resulting dashboard provides two kinds of insights into content filter effectiveness. The first focuses on Content Detection with visualized insights into metrics like Total blocked request count and block rate, Blocked requests by category, Severity distribution by category and more. The second focuses on Abusive User Detection to highlight how regularly the content filters safeguards are abused by end users and identify the severity and frequency of those occurrences.

🔖 | Learn more in this post
🔖 | Review the main announcement

8 | Get Started Exploring

This was a lot - and it is still just the tip of the iceberg when it comes to actively understanding and applying responsible AI principles in practice. Want to get started exploring this topic furthere? Bookmark and revisit these three core resources:

Resource	Description
1️⃣ Blog Post	Official Announcement of New Tools
2️⃣ Collection	My Responible AI For Developers Collection
3️⃣ Collection	My Azure AI For Developers Collection

Train & Debug ML Models for Responsible AI - Join the #AISkillsChallenge!

Nitya Narasimhan, Ph.D — Sat, 23 Mar 2024 22:49:30 +0000

Welcome to the eleventh post in my This Week In AI News series. Want to keep up with my weekly posts? Now there's a tag you can follow: 👇🏽

#thisweekinai Follow

The 3 Resources To Know This Week

Resource	Description
1️⃣ Registration	Microsoft Learn AI Skills Challenge
2️⃣ Collection	Responsible AI Resources For Developers
3️⃣ Documentation	The Responsible AI Toolbox Website

✨✨ Microsoft Learn AI Skills Challenge has started! ✨✨

Whether you are an AI startup founder ideating your business model, a data scientist working with LLMs or a software developer building a RAG solution, we got you covered! Immerse yourself in a skilling journey combining self-study (with Cloud Skills Challenge) with instructor-guided livestream sessions (Learn Live with Subject Matter Experts):
🚀 | Learn More: Learn Live Series
🚨 | Register For My Session: April 10, 2024

Earlier in this series, I shared this post which focused on three things with respect to Responsible AI:

The 6 guiding principles for Responsible AI
The Responsible AI Developer Hub for self-guided learning.
3 Responsible AI Workshops for hands-on skilling.

Responsible AI For Developers: Resources For Self-Guided Learning

Nitya Narasimhan, Ph.D for Microsoft Azure ・ Jan 31

#azure #beginners #ai #thisweekinai

Today, I want to dive deeper into one specific workshop and tell you about an ongoing AI Skills Challenge where we will be running a live training session (online, free) where you can get hands-on experience with the Responsible AI Dashboard guided by us! But first, let's talk about the AI Skills Challenge and why it matters.

1 | The Microsoft Learn #AISkillsChallenge

The Microsoft Learn #AISkillsChallenge is a 9-part series of livestreamed training sessions that cover the gamut of topics from generative AI to LLM Ops. Visit this link for registration pages and details for each episode - or view the tweet below for a quick video tour.

// Detect dark theme var iframe = document.getElementById('tweet-1771638148679815478-269'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1771638148679815478&theme=dark" }

2 | Register to Learn Live

The Responsible AI episode is scheduled for April 10 - just click the link below to go directly to the registration page for this event.

🚨 | Register Here: April 10, 2024

Then join us on the livestream for a hands-on walkthrough of this Microsoft Learn module. All you need is your laptop and a modern browser - the module comes with an Azure sandbox so we can dive straight into learning by doing, with no other setup overheads.

The session will be 90-minutes long, with time for live Q&A. While we do recommend following along with us, you should be able to continue working on the training even after the livestream is done. Read on for more details on what you will learn, and the process involved.

3 | What We'll Be Doing

We'll learn how to train a machine learning model on Azure, then debug it to ensure that it applies responsible AI principles in practice.

We'll build our model based on the UCI Diabetes dataset which has a decade of data from 130 US hospitals, with 50+ features recorded including attributes like patient number, race, gender, age, admission type, time in hospital, number of medications, number of outpatient, inpatient, and emergency visits in the year before hospitalization, etc.

Our model can then predict the likelihood of a patient needing to be readmitted to hospital within 30 days - allowing a physician or hospital to them make decisions related to patient care.

4 | Why Debug Models?

The reality is that model predictions can have real-world implications when used for decision-making. Models are rarely 100% accurate - so humans (doctors, admins) need to make their decisions based on metrics like accuracy and confidence. Poor prediction can result in sick people being denied critical care, or healthy ones being billed for unnecessary hospitalizations.

With the Responsible AI Dashboard, we can learn to debug models interactively using different components (error analysis, model performance, data biases, model interpretability) that can be analyzed across different cohorts (subsets of data matching a common set of attributes) to ensure that predictions are made fairly, correctly, and consistently for everyone.

5 | How We Debug Models

The training session consists of two main sections, each with code and documentation in the form of Jupyter Notebooks for convenience.

In the first section, we'll learn to setup an Azure Machine Learning Studio workspace, upload our test and training data, setup a training job, and register our trained model on Azure to expose a prediction endpoint for applications.
In the next section, we'll create a Responsible AI Dashboard for the deployed model on Azure, selecting the required subset of components from our Responsible AI toolbox. The result is an interactive dashboard that allows us to create cohorts based on various criteria (e.g., by error rates) and use them to analyze the model for fairness, performance and other responsible AI practices.

6 | What is the Responsible AI Toolbox?

Some of you may have heard of The Responsible AI Toolbox repository maintained by Microsoft Research, and wondered how that relates to the Responsible AI Dashboard.

The toolbox refers to the suite of components for operationalizing Responsible AI end-to-end, many of which are open-source projects from Microsoft or the community. See the figure below to understand the components involved.

The dashboard is a single pane of glass (unified interface across tools) that helps you easily flow through different stages of model debugging (identification, diagnosis, and mitigation) and decision-making. See the figure below to get a sense of how the toolbox components are organized into dashboard views for model debugging (left) and decision making (right).

There is a lot more to learn about Responsible AI in the context of end-to-end development workflows for predictive AI and generative AI applications. By joining us in this training, you should be able to jumpstart your learning journey with a hands-on exercise that can help translate abstract concepts into real-world practice.

🚨 | Register Here to join us: April 10, 2024

7 | Related Resources

Here are the three main resources for this post:

Resource	Description
1️⃣ Registration	Microsoft Learn AI Skills Challenge
2️⃣ Collection	Responsible AI Resources For Developers
3️⃣ Documentation	The Responsible AI Toolbox Website

And don't forget to follow the Microsoft Azure publication or the individual author pages, for more updates.

Microsoft Azure Follow

Learn how to build and manage powerful applications using Microsoft Azure cloud services. Explore the documentation and samples, skill yourself up with tutorials and training. Any language. Any platform.

Nitya Narasimhan, Ph.DFollow

PhD & Polyglot - AI Advocate @Microsoft - Community Builder @NY - Innovator (12+ patents) - Speaker & Keynoter - ex-Researcher (UbiComp, DistSys) - Visual Storyteller @SketchTheDocs - Proud Parent

Ruth YakubuFollow

Ruth Yakubu is a Principle Cloud Advocate at Microsoft. Ruth specializes in Java, Advanced Analytics, Data Platforms and Artificial Intelligence (AI). In addition, she's been a tech speaker at severa

Python Data Analysis with Developer Tools & AI (and #14DaysOfDataScience)

Nitya Narasimhan, Ph.D — Sat, 16 Mar 2024 14:35:29 +0000

Three Resources To Know This Week

Resource	Description
Collection 1️⃣	Skill up on Data Science Tools & Techniques
Collection 2️⃣	Skill up on Responsible AI Principles & Tooling
Collection 3️⃣	Build Generative AI Apps End-to-End with Azure AI

Welcome to the tenth post in my This Week In AI News series. Want to keep up with my weekly posts? Now there's a tag you can follow: 👇🏽

#thisweekinai Follow

In my previous posts, I've talked about LLM Ops and the application lifecyle for building generative Read my tech community post on the topic for a quick refresher and core sample repos. Then, let's shift left even further, and think about where it all starts -- with data science and analysis. Here's what we cover in this post:

1 | Python Data Science Day
2 | Python Data Analysis Workshop
3 | Week 1 - A Focus on Fundamentals
4 | Week 2 - A Focus on Developer Tools
5 | Self-Guided Learning Resources

1 | Python Data Science Day

March 14, 2024 (3.14) was Pi Day aka Python Data Science Day at Microsoft - a full-day of talks from Python enthusiasts and experts from all over the world. Watch the replay of the 8-hour livestream below, and check the description for the complete list of talks with timestamped links into the stream for quick access.

2 | Python Data Analysis Workshop

I presented a talk on Simplifying Data Analysis with Developer Tools & AI targeting the non-Python developer. My target audience was someone new to Python or Data Science, but otherwise experienced in development. And my goal was to provide a learning roadmap and quickstart environment so they could get productive quickly in their data science journey.

Find the talk video at this timestamp in the livestream.
Browse the talk slides here:

In that talk, I outlined the following roadmap for developers new to this topic, to structure their learning journey but also create a reusable, shareable and reproducible development environment for producitivity. And I share a workshop repo I am maintaining, that can take developers from conceptual understanding to hands-on practice. Want to learn more? Check out the Week 2 section below.

But wait - what if you wanted to have a more structured roadmap of what to learn, and where to look for resources to learn it? We have you covered! Say hello to #14DaysOfDataScience - a series that bookends the Data Science Day event with 2 weeks of daily blog posts focused on Data Science & Developer tools authored by Data & AI Advocates at Microsoft.

3 | Week 1 - A Focus on Fundamentals

The first week of the series focused on Data Science Foundations - from the definition of data science, to understanding the data science application lifecycle end-to-end. Along the way, you'll also get an introduction to machine learning (supervised and unsupervised) and responsible AI, and understand what the data science developer experience involves.

Check out the series today, starting with this first post which also contains a series listing that makes it easy to find the other posts, including upcoming ones in Week 2.

Welcome to 14 days of Data Science!

Renee Noble for Microsoft Azure ・ Mar 8

#datascience #machinelearning #python

Want to learn more about data science fundamentals? Follow these Week 1 authors right here on dev.to:

Jasmine GreenawayFollow

Brooklyn based programmer 🗽 Cloud Advocate by day, instructor and meetup organizer by night.

Renee NobleFollow

I'm passionate about technology, education, and community. I work to combine these in many ways through many jobs!

4 | Week 2 - A Focus on Developer Tools

The second week of the series focuses on Developer Tools taking you from concepts (week 1) to code (week 2) with hands-on exercises to solidify what you learn.

In the earlier section on Python Data Analysis (my Data Science Day talk), I shared a roadmap with a link to a workshop repo that will be a source for these hands-on exercises. This week, I will be updating that repo and publishing one post each day, to showcase *one developer tool at a time.

The series kicked off yesterday with this post that helps set the stage with a look at Dev Containers & GitHub Codespaces. By the end of this post, you should have forked the repo and validated that you have a working development environment with Jupyter Notebook support.

Data Science & DevTools: GitHub Codespaces

Renee Noble for Microsoft Azure ・ Mar 15

#beginners #datascience #githubcodespaces #developers

We will then build on that project throughout the week, to explore more tools and concepts, one post at a time. Check back for the updated list of posts here 👇🏽:

Data Science & DevTools: Visual Studio Code

Renee Noble for Microsoft Azure ・ Mar 17

#datascience #beginners #developers #vscode

5 | Self-Guided Learning Resources

Bookmark and revisit these for more information:

Resource	Description
Collection 1️⃣	Skill up on Data Science Tools & Techniques
Collection 2️⃣	Skill up on Responsible AI Principles & Tooling
Collection 3️⃣	Build Generative AI Apps End-to-End with Azure AI

And don't forget to follow this tag for updates from This Week in AI News:

#thisweekinai Follow

And follow our Azure publication right here on dev.to for more articles from the Cloud Advocacy and Product Teams at Microsoft:

Microsoft Azure Follow

Fuel Your Intelligent Apps with Azure AI

Nitya Narasimhan, Ph.D — Tue, 12 Mar 2024 11:50:27 +0000

Three Resources To Know:

Welcome to the ninth post in my This Week In AI News series. Want to keep up with my weekly posts? Now there's a tag you can follow: 👇🏽

#thisweekinai Follow

Introduction

In my previous post, I talked about how you can build a copilot experience using Azure AI Studio at a very high level.

Building a Copilot Code-First with Azure AI

Nitya Narasimhan, Ph.D for Microsoft Azure ・ Mar 7

#azure #beginners #ai #thisweekinai

By the end of that post, you should have been able to answer three questions:

What is LLM Ops? (paradigm shift)
What is Azure AI Studio? (unified platform)
What is a Copilot? (generative AI app with your data)

The post also linked to a video that walked through the process of building a copilot code-first. But what if you wanted a more hands-on walkthrough of the process with an end-to-end application sample or quickstart templates for popular frameworks? We have you covered!

This week, Azure AI takes the spotlight on the #60DaysOfIA series covering core tools and technologies to build intelligent apps using cloud-native technologies on Azure.

Here's what you can expect to learn:

Day 1️⃣ | End-to-End Contoso Chat Sample

Day 1 kicks off with a 2-part post:

A kickoff post that describes the driving application scenario, the paradigm shift to LLM Ops, and the "copilot" application experience on Azure.

A deep-dive into Contoso Chat and end-to-end application sample that teaches you how to build, evaluate, deploy, and test, a RAG-based chat AI application using Azure AI Studio and Prompt flow.

Days 2️⃣-4️⃣: Copilot Quickstart Samples

The figure below shows a high-level architecture diagram for a basic "copilot" application. In this context, a copilot is a generative AI application grounded in your data that allows the user to perform complex tasks or ask questions using conversational chat interface.

Building a copilot requires us to think about two main things:

Chat Function - developing the chat function that coordinates or orchestrates the many interactions required, to implement the Retrieval Augmented Generation (RAG) pattern shown.
Chat API - deploying the chat function to a hosted endpoint to expose an API that you can interact with from a chat UI (e.g., Contoso Web application) to deliver conversational user experiences with your products.

In this part of the series, we'll explore three quickstart samples that can jumpstart your own development thinking, using basic Python code (with the Azure AI SDK), or by integrating orchestration tools and popular frameworks like Prompt flow and LangChain

Day 5️⃣: Deploy AI Responsibly

The previous posts focus on Azure Samples that help you build a copilot application on the Azure AI platform with a focus on ideation and augmentation steps of the LLM Ops workflow. They end with deployment to Azure, to expose an API endpoint for integration.

In this final post of our journey we circle back to the first post and revisit the operationalization phase of the LLM Ops story with a focus on 2 aspects:

Deployment Options - How can we test the deployed endpoint on Azure AI Studio? How can we integrate it with our applications? And are there enterprise-grade examples that show how to containerize and deploy such solutions with cloud-native technologies?

Responsible AI - How can we ensure that we bring responsible AI practices into the ideation, augmentation and operationalization phases in a meaningful and effective way for generative AI applications? We explore the steps of identifying potential harms, evaluating solutions to assess quality and responsible AI usage, and applying content filters for content safety in production.

Ask The Expert (Mar 21)

Read the series and now you have questions or feedback? Or perhaps you built something interesting and want to share your insights or learn what's next to help you evolve your idea?

We have you covered! Don't forget to join us Mar 21 for an #AskTheExpert where authors from this series will be live to take your questions and share their perspectives on the generative AI development journey.

🚨 Register Now to attend.

Three Resources To Know:

Want to get the latest news, updates, and content on building intelligent apps with Azure? Follow the Azure org right here on dev.to - and subscribe to my This Week in AI series by following the #thisweekinai tag.

Microsoft Azure Follow

#thisweekinai Follow

Building a Copilot Code-First with Azure AI

Nitya Narasimhan, Ph.D — Thu, 07 Mar 2024 16:10:10 +0000

3 Resources To JumpStart Your Azure AI Studio Exploration:

Welcome to the eighth post in my This Week In AI News series. Today, I want to talk about building generative AI applications code-first with the Azure AI platform. What does this mean and how does it help us make the paradigm shift to LLM Ops?

What is LLM Ops?

Traditionally, when we talk about AI applications we were referring to machine learning solutions where we built custom models trained on relatively-finite datasets for our target use case. The deployed AI application would then be used to make predictions.

Today, when we talk about AI applications we typically mean generative AI solutions based on large language models trained on massive datasets. Our focus is then on prompt engineering and fine-tuning these pre-trained models to generate content.

This has led to a paradigm shift from ML Ops to LLM Ops where the end-to-end application development lifecycle looks more like the picture shown below. Building these solutions now requires us to think about 3 phases:

Ideation - define and build the basic experience
Augmentation - evaluate and refine for quality
Operationalization - deploy and use in production

This requires new tools, technologies and processes to streamline developer experiencea in building, testing, deploying & integrating, these apps.

What is Azure AI Studio?

Azure AI Studio tackles this challenge by providing a unified platform that supports the entire workflow from ideation (explore models, engineer prompts) to augmentation (build and manage AI projects) to operationalization (deploy & monitor solutions).

Want to get started exploring the platform? Check out the Azure AI Studio UI - a browser-based experience perfect for low-code developers. But if you're a professional developer, you probably want to have more control over the interactions, and potentially integrate additional libraries or features to enhance your solution. This is where having support for code-first development helps.

What does code-first development mean?

From the Azure AI Studio perspective, this means supporting command-line (CLI) and programmatic (SDK) interactions with the underlying Azure AI platform and resources. In February, the Azure AI team recorded this livestream talk which walks you through the process of building an enterprise copilot AI experience using a code-first approach on Azure AI Studio.

Want a more complete picture of what is involved? Check out my Tech Community Post from last month for a more detailed description of the tools, process, and resources to skill up on this topic. The tweet below has a preview of the post for convenience.

// Detect dark theme var iframe = document.getElementById('tweet-1765756581709312147-42'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1765756581709312147&theme=dark" }

Azure AI Week & AskTheExpert

But there's more. Keep an eye out on Mar 11-15 (next week) as we launch Azure AI Week on #60DaysOfIA - part of a multi-week campaign with events and activities focused on building intelligent apps.

Then join us on March 21 for an #AskTheExpert session we will take your questions live, share demos and discuss the generative AI developer journey!

🚨 Register Now to attend.

This is an exciting time for developers, data scientists and entrepreneurs to go from ideation to operationalization and build intelligent generative AI experiences code-first! Don't forget to check out the resources below to start skilling up!

3 Resources To JumpStart Your Azure AI Studio Exploration:

Fine-Tuning Fundamentals - Generative AI For Beginners (v2)

Nitya Narasimhan, Ph.D — Fri, 01 Mar 2024 10:43:41 +0000

3 Resources to Jumpstart Your Generative AI Journey:

Welcome to the seventh post in my This Week in AI News series! Today, I want to continue my post on Generative AI For Beginners and move on from prompt engineering to the related topic of fine tuning. Let's dive in!

Generative AI For Beginners (v2)

In the previous post, I covered the v1 edition of this open-source curriculum (chapters 1-12) released in Oct 2023 - with specific focus on "Prompt Engineering Fundamentals", the chapter I contributed.

Prompt Engineering Fundamentals - Generative AI For Beginners (v1)

Nitya Narasimhan, Ph.D for Microsoft Azure ・ Mar 1

#opensource #ai #beginners #thisweekinai

In today's post, I'll dive into the v2 extension to the curriculum that was just released in Feb 2024 (chapters 13-18) - and focus in more detail on the "Fine Tuning" chapter I contributed. But first, let's take a look at what's new in v2:

Lesson	Description
Securing Generative AI	Covers the adversarial threat landscape for AI and looks at options for security testing, data protection, and safety evaluation (including red teaming)
Generative AI App Lifecycle	covers the paradigm shift from MLOps to LLMOps - and explores the workflow and tools to streamline end-to-end development
Retrieval Augmented Generation	Covers a core technique to improve LLM response quality by grounding it in your own data and using embeddings and vector databases.
Open-Source Models	Covers benefits of open-source models like Llama2, Mistral and Falcon - and the value of model hubs like Hugging Face for discovery & integration.
AI Agent Systems	Covers evolution of AI apps from assistants (interactive, chat) to agents (autonomous, task execution) e.g., AutoGen, LangChain Agents & JARVIS.
Fine-Tuning Models	Covers ability to retrain foundation models with new examples to improve response quality or reduce usage costs & complexity for prompt engineering.

Prompt Engineering To Fine Tuning

In the v1 edition, I talked about the value of prompt engineering to improve the quality of model responses to user questions. Specifically, we looked at ways to construct the prompt using techniques like few-shot learning, prompt templates, system prompts and more - all of which enhance the default user prompt with additional content or context to guide the LLM towards more relevant responses.

But prompt engineering alone may not be enough:

Cost: Models have tokenization limits and usage costs that can constrain the degree to which you can enhance the default prompt. This limits the number of examples you can add in primary content, or the richness of responses in completions.
Customization: You may want to add new skills or capabilities to the model that enhance its behavior across user interactions. Doing this on a per-prompt basis is inefficient and may not even be possible.

So, how can you take advantage of the rich language models available to you, while getting the cost-effective and customized user experiences you need? This is where fine-tuning models can help.

Fine-Tuning Fundamentals

In lesson 18 of this curriculum, we tackle this challenge by learning about Fine-Tuning Models using new data or examples. The lesson and covers the following topics at a high-level, providing related resources and a hands-on assignment for self-guided deeper dives. Check out the illustrated guide below for more detail on what each topic covers.

Sub-Topic	Description
Introduction	Understand foundation models and related concepts for response quality
Motivation	Learn why fine-tuning matters and when to start exploring it as an option
Process	Understand the steps in a fine-tuning workflow, and the related challenges
Data Prep	Ensure you have the right data quantity and quality for fine-tuning
Training	Run the fine-tuning job, monitor progress, then test & iterate for quality
Deployment	Make your fine-tuned model available for real-world interactions, know the constraints

Fine-tuning is a fascinating topic with great potential for experimentation and learning. Expect more updates to that lesson with focus on walkthroughs of assignments that show the application of these concepts to real-world use cases and models. For now, start by exploring the resources for self-guided learning provided in that chapter.

Summary & Next Steps

In this post, we looked at what the v2 edition of the Generative AI for Beginners curriculum provides, with some focus on the Fine Tuning Fundamentals chapter that concludes it.

This is a fast-evolving space so expect more updates to the curriculum - your feedback and contributions are welcome! Start your journey today by forking the repo to your profile and exploring the lessons at your own pace!!

Happy learning!

3 Resources to Jumpstart Your Generative AI Journey: