Forem: Arsh Tariq

DJ Electrohead: Your AI-powered DJ simulator

Arsh Tariq — Sun, 11 May 2025 21:37:52 +0000

This is a submission for the Amazon Q Developer "Quack The Code" Challenge: That's Entertainment!

What I Built

DJ Electrohead is an AI- powered interactive DJ simulation game that combines music mixing with crowd interaction. The game features three distinct play modes:

Crowdplay Mode: Connect with Spotify to mix real tracks. An AI-powered crowd reacts to your song choices, analyzing how well you maintain energy and flow between tracks.

Strudel Mode: Create live beats with code using the strudel repl's musical loops and patterns. The crowd responds to your rhythm and creativity in real-time.

Mixing Mode: Create live mixes of two songs alongside the multitude of tools you would find on a real DJ's turntable!

Key features:

Dynamic crowd reactions with a real-time vibe meter powered by deepseek-VL v3
Score system with combo multipliers for maintaining crowd energy
Animated DJ character sprites that responds to the crowd's mood depending on a dynamic crowd vibe meter
High score tracking across both game modes
Seamless Spotify integration for authentic DJ experience and the ability to play any song you want!
Strudel allows users to flex their coding ability whilst creating banger music
Users can actually DJ using turntables along with snares, kicks, pitch changes, reverbs and flangers!

Demo

Demo Video

Code Repository

Repository Link

How I Used Amazon Q Developer

I first thought about what I wanted to build and then asked chatgpt to create a text document titled: "Tasks.txt" (available in the github repository)
Then I instructed Amazon Q developer to use the file tasks.txt as a starting point to get a feel for what it has to build.
Since the task sheet is split into different phases, I made Q developer stop everytime it completed a major task in each phase and tweaked the code manually or using q dev itself if needed.
After reviewing every phase's features that q developer implemented, I pushed it to my repository and moved on to the next phase.

GLaDOS - Your personalized AI Game Recommender

Arsh Tariq — Sat, 25 Jan 2025 10:57:46 +0000

*This is a submission for the Agent.ai Challenge:

Full-Stack Agent (See Details)*

Assembly of Agents (See Details)

Productivity-Pro Agent (See Details)*

What I Built

I built GLaDOS, an AI-powered game recommendation system that recommends games and links to buy them based on user's favorite games.

Why I Built This

Gamers are always looking to play more titles that are familiar enough to them genre wise but are entirely new adventures, yet they have no idea what to play. GLaDOS (inspired by the iconic character from Portal) is here to solve that issue by providing tailored recommendations to its users.

Demo

Check out GLaDOS in action:

Demo Link

Features

Dynamic User Input: Users simply provide their steam profile and GLaDOS does the rest.
Custom Game Recommendations: GLaDOS creates recommendations tailored to what you have already played and enjoyed.
Personalized Suggestions: Recommend games based on user taste, past played games, and current favorite games (most hours spent).

How I Built It

1. Web Scraping: Used agent.ai builder's to scrape a user's steam profile when provided with the link.

2. Information extraction: Using GPT-4o to extract necessary and relevant information only from the web scraping.

3. AI-Driven Recommendations: GLaDOS takes into account past played games, games in showcase and favorite games to generate recommendations.

Agent.ai Experience

Building GLaDOS with Agent.ai was both rewarding and challenging:

Delightful Moments

The no-code builder made this whole process seamless and easy.
Watching GLaDOS craft amazing and personalized recommendations felt like magic!
The intuitive design of the agent AI builder made it very easy to use

Challenges

Extracting the proper relevant data from the steam profile.
Fine-tuning the AI responses for properly generating the required results.
Fine-tuning the prompts and engineering a solution to get the best possible results.

Future Enhancements

Bigger Library: Gaining access to the user's other accounts across platforms and recommending games via those.
Mobile App Development: Enable users to access recommendations on the go and also implement this for mobile games.
Real-Time Feedback: Incorporate real-time user feedback to improve game recommendations.
Multi-Language Support: Make the app accessible to a global audience.
Recommendation History: Track past recs and suggest newer ones based on what the user liked before.

Thank you for exploring GLaDOS, your AI-powered game recommender! Do share your love and support!

Chef HAL-9000: Cooking Up Masterpieces, One Recipe at a Time!

Arsh Tariq — Fri, 24 Jan 2025 20:52:04 +0000

*This is a submission for the Agent.ai Challenge:

Full-Stack Agent (See Details)*

Assembly of Agents (See Details)

Productivity-Pro Agent (See Details)*

What I Built

I built Chef HAL-9000, an AI-powered culinary assistant designed to help users craft personalized, delicious recipes based on their available ingredients, preferences, and dietary needs.

Why I Built This

Cooking can be a time-consuming and daunting task, especially when you're trying to balance flavor with dietary restrictions. Inspired by HAL-9000 from 2001: A Space Odyssey, Chef HAL-9000 takes the guesswork out of meal planning by generating personalized recipes that are practical, easy to follow, and tailored to your tastes and ingredients.

Chef HAL-9000 is perfect for:

Home cooks who want quick, creative meal ideas.
Busy individuals looking to streamline their meal prep.
Anyone trying to stay on track with specific dietary goals or restrictions.

Demo

Check out Chef HAL-9000 in action:

Demo Link

Video Demo:

YouTube scraper

Screenshots and Workflow

1. Input for Chef HAL-9000

Describe your ingredients, and let Chef HAL-9000 do the rest!

2. Chef HAL-9000 Actions

The workflow for how Chef HAL-9000 generates recipes.

3. Generated Recipe

The recipe that Chef HAL-9000 creates based on your input.

4. Specific Recipe Mode

Users can choose the find recipes for specific dishes along with visual tutorials for guidance.

5. Ingredient Specific Generation

The Ingredient specific mode, which generates recipes using only what is
readily available at home.

Features

Dynamic User Input: Users simply provide their ingredients, preferences, and any dietary restrictions.
Custom Recipe Generation: AI creates recipes tailored to what you have on hand.
Real-Time Ingredient Substitutions: Suggests ingredient substitutions based on availability or dietary preferences.
Day-by-Day Meal Planning: Detailed, actionable plans for meal prep.
Budget-Friendly: Recipes are generated with cost-effective options when necessary.
Personalized Suggestions: Recommend recipes based on user taste, past meals, and preferences.

Use Cases

Home Cooks: Discover creative ways to use everyday ingredients.
Busy Professionals: Generate quick, easy recipes for meal planning.
Health-Conscious Eaters: Tailored meals based on specific dietary needs like vegan, gluten-free, or keto.
Families: Plan family meals with simple, budget-friendly recipes.
Food Enthusiasts: Explore new and exciting recipes from around the world.

How I Built It

1. If/Else directives: Used agent.ai builder's if/else statements to include a dropdown option and let users decide what they want to do.

2. Recipe Generation: Using GPT-4o to generate recipes based on the available ingredients and dietary preferences (if needed).

3. AI-Driven Tutorial: Implementing a YouTube web scraper to find relevant tutorials to act as a visual guide.

4. Multi-Agent Workflow: Connecting agents for recipe generation and tutorial finding together.

Agent.ai Experience

Building Chef HAL-9000 with Agent.ai was both rewarding and challenging:

Delightful Moments

The no-code builder made integrating multiple agents seamless.
Watching Chef HAL-9000 craft delicious and personalized recipes felt like magic!
The multi-agent system made it easy to connect various features like ingredient searching and recipe generation.
The intuitive if/else statements helped me a lot in figuring out how to set the whole agent up. It felt very rewarding to complete this agent.

Challenges

Balancing recipe creativity with ingredient limitations.
Fine-tuning the AI responses for diverse dietary needs and preferences.
Fine-tuning the prompts and engineering a solution to get the best possible results.

Future Enhancements

Collaborative Meal Planning: Allow users to plan meals together in a shared platform.
Mobile App Development: Enable users to access recipes on the go.
Real-Time Feedback: Incorporate real-time user feedback to improve recipe recommendations.
Multi-Language Support: Make the app accessible to a global audience.
Recipe History: Track past meals and suggest recipes based on what you've cooked before.

- Conversational AI: Make the agent fully conversational and also implement voice communication

Thank you for exploring Chef HAL-9000, your AI-powered culinary assistant! Do share your love and support!🍴

Tyche: Crafting Adventures, One Surprise at a Time!

Arsh Tariq — Wed, 22 Jan 2025 19:28:03 +0000

This is a submission for the Agent.ai Challenge: Full-Stack Agent (See Details)

Assembly of Agents (See Details)

What I Built

I built Tyche, an AI-powered travel planning agent that transforms user inputs into spontaneous, exciting, and personalized travel plans.

Why I Built This

Travel planning often involves extensive research, and spontaneity is hard to balance with budgets and schedules. Tyche, inspired by the Greek goddess of fortune and chance, takes the guesswork out of trip planning by generating a complete, random travel itinerary tailored to the user’s preferences, including destination, activities, and budget.

Tyche is perfect for:

Adventurers who want a surprise element in their travels.
Busy individuals seeking a streamlined way to plan trips.
Anyone looking to make travel planning stress-free and fun.

Demo

Check out Tyche in action:

Demo Link

Video Demo:

Flight Planner Agent
Travel researcher Agent
Webservice Repo

Screenshots and Workflow

1. Input for Tyche

Describe your ideal trip, and let Tyche take over!

2. Tyche Actions

The workflow for the main agent.

3. Generated Itinerary

The itinerary that the agent created for the user based on their prompt.

4. Travel Researcher Agent

The Travel Researcher agent invoked by the main agent.

5. Flight Planner Agent

The Flight Planner agent invoked by the main agent, which pings a web service I created using the FlightRadar24 API and hosted on Render.

Using Cron-Job to ping the web service every 15 minutes to keep it awake (free tier limitations 😞).

Features

Dynamic User Input: Users describe their dream trip, including preferences, budget, and duration.
URL-Based Content Generation: AI generates content based on landing pages for destination insights.
SEO Optimization: Enhances blog posts to rank better in search results.
Platform-Specific Outputs: Creates content optimized for LinkedIn, Twitter, and blogs, perfect for promoting travel ideas.
Real-Time Web Integration: Scrapes live travel data to find the best deals for flights, accommodations, and activities.
Day-by-Day Planning: Clear, actionable plans with links to book directly.
Budget Management: Keeps expenses within user-specified limits with a detailed cost breakdown.
High-Quality Outputs: Generates professional, polished, and contextually relevant content for all touchpoints.

Use Cases

Adventurers: Plan spontaneous trips with ease.
Founders: Create trip-related content to build personal or business brands.
Marketing Teams: Align travel content with the company’s voice.
Social Media Managers: Schedule optimized LinkedIn and Twitter posts effortlessly.
Travel Enthusiasts: Enhance online visibility, traffic, and conversion rates.

How I Built It

1. Data Scraping: Leveraged web scraping by utilizing another agent to collect structured data from URLs for destination-specific insights.

2. AI Analysis: Processed extracted data to understand travel details like problems, solutions, and target audiences.

3. Generative AI Model: Used advanced NLP techniques to generate high-quality content tailored to the scraped data.

4. Fine-Tuning: Ensured the tone, style, and platform-specific engagement were optimized.

5. Multi-Agent Workflow: Connected agents for planning, research, and booking into a seamless pipeline.

6. Flight search API: Created a webservice for searching flights and feeding back into the agent.

Agent.ai Experience

Building Tyche with Agent.ai was both rewarding and educational:

Delightful Moments

The no-code builder simplified complex integrations.
Chaining multiple agents into one cohesive flow was a game-changer.
Watching Tyche generate highly engaging itineraries felt magical!

Challenges

Managing large outputs while ensuring user-friendly clarity.
Fine-tuning agent responses to balance spontaneity with constraints.
Addressing free-tier limitations for always-on web services.

Future Enhancements

Collaborative Trip Planning: Invite friends to co-plan trips.
Mobile App Development: Enable users to plan on the go.
Data-Driven Insights: Integrate detailed analytics for continuous improvement.
Multi-Language Support: Broaden accessibility for global users.
Feedback Integration: Iterate itineraries based on user preferences.

Thank you for exploring Tyche, your AI-powered gateway to adventure! 🚀