Forem: Omer Ben Ami

What I've Learned About Deploying ML & LLM Projects from DAGsHub's CEO

Omer Ben Ami — Wed, 19 Mar 2025 17:46:26 +0000

Hey guys!
I've been experimenting with building and deploying ML and LLM projects for a while now, and honestly, it’s been a journey.

Training the models always felt more straightforward, but deploying them smoothly into production turned out to be a whole new beast.

Recently, I had a really good conversation with Dean Pleban (CEO @ DAGsHub), who shared some incredibly practical insights based on his own experience helping teams go from experiments to real-world production. Here are a few valuable lessons Dean shared with me that I think could help others as well:

Data matters way more than I thought.
Initially, I focused a lot on model architectures and less on the quality of my data pipelines. Production performance heavily depends on robust data handling—things like proper data versioning, monitoring, and governance can save you a lot of headaches. This becomes way more important when your toy-project becomes a collaborative project with others.
LLMs need their own rules.
Working with large language models introduced challenges I wasn't fully prepared for—like hallucinations, biases, and the resource demands. Dean suggested frameworks like RAES (Robustness, Alignment, Efficiency, Safety) to help tackle these issues, and it’s something I’m actively trying out now. He also mentioned "LLM as a judge" which seems to be a concept that is getting a lot of attention recently.

Some practical tips Dean shared with me:

Save chain of thought output - you never know when you might need it.
Log experiments thoroughly (parameters, hyper-parameters, models used, data-versioning...).
Start with a Jupyter notebook, but move to production-grade tooling (all tools mentioned in the guide bellow 👇🏻)

To help myself (and hopefully others) visualize and internalize these lessons, I created an interactive guide that breaks down how successful ML/LLM projects are structured. If you're curious, you can explore it here:

https://www.readyforagents.com/resources/llm-projects-structure

I’m still learning a ton, so I'd genuinely appreciate hearing about your experiences too—what’s been the trickiest part of deploying ML & LLM projects for you?

Looking forward to your insights!

I gem-packed this with how I'm leveraging LLMs in my workflow!

Omer Ben Ami — Sun, 12 Jan 2025 20:47:42 +0000

Supercharge Your Skills: A 2025 Guide for Developers

Omer Ben Ami ・ Jan 12

#webdev #productivity #programming

Supercharge Your Skills: A 2025 Guide for Developers

Omer Ben Ami — Sun, 12 Jan 2025 18:00:36 +0000

Hey there, devs! 👋

As a product builder working on a SaaS app, I've discovered some techniques that have turbocharged my development process.

These are hacks most devs still aren't using! so you can get ahead by consistently working smarter. For full details check this video

My go-tos:

The basics

Screenshot your errors: When you hit a bug, snap a screenshot and feed it to your AI assistant. Ask "Why am I getting this?" for quick insights.
Treat AI like a colleague: If the initial solution doesn't cut it, keep the conversation going with your AI assistant, just as you would with a human teammate.

Boost Your Learning and Retention - Here's where the magic happens:

Set up custom text replacements: Create shortcuts like "eli5" and "eli6" on your Mac for quick access to learning prompts.
Use the "ELI5" approach: Before implementing a solution, ask the AI to "Explain it like I'm a junior software developer" for a comprehensive breakdown.
Request analogies: Use the "ELI6" prompt to get explanations with analogies and simple examples. It makes complex concepts stick!

Build Your Personal Knowledge Base

This is crucial for long-term growth:
Create a public GitHub repo: Set up a separate repository for your project documentation.

Document your learnings: After solving a problem, create a markdown file explaining the concept. Include AI-generated charts and UML diagrams for visual learners - I've created a full repo with all the details I might need in the future autogenerated:

By implementing these strategies, you'll not only solve problems faster but also create a valuable resource for your future projects.

It's all about continuous learning and improvement.

What do you think, dev fam?
I'm curious - what has been useful for you to get the most out of your AI-editor?

Building in public a SaaS for $9K

Omer Ben Ami — Wed, 11 Sep 2024 16:10:03 +0000

Hi friends 👋

I'm currently working on building a SaaS platform to help Pilates studios manage personal training sessions - I'm charging 9k$ and I'll be sharing everything.

My promise to you is - any question will be answered!

Why Pilates Studios?

To be honest, this opportunity found me. kind of.
The story goes like so - I was building a board to put together all yoga sessions around me. While I was working on it in a local cafe, I overheard two yoga teachers chatting about their next session next to me. I immediately thought to myself - "a chance for feedback!". So I showed it to them, it wasn't even live, just ran it on my laptop.

One thing led to another and one of them asked me to build this app :)

Tech Stack

Vite: Lightning-fast bundling and development environment 🚀
Shadcn: UI components to speed up frontend development 🖌️
Firebase: For user authentication, database, and backend services 🛠️

To be honest, this is the first production-grade product I'm building from scratch so cross your fingers for me!

Build in Public

I'll be totally transparent about the tech, resources and finances, so if you're curious about the breakdown, I'll share the full cost and details over twitter.

Interested? Check this tweet and buckle up for the ride!

// Detect dark theme var iframe = document.getElementById('tweet-1833420056727818652-598'); if (document.body.className.includes('dark-theme')) { iframe.src = "https://platform.twitter.com/embed/Tweet.html?id=1833420056727818652&theme=dark" }

Noob builds a SaaS in public so you all can learn too

Omer Ben Ami — Fri, 19 Jan 2024 11:06:46 +0000

Hi friends 👋

If you ever felt like the product manager working with you knows nothing about how hard your work is, then, you might be right!

That's why as a PM I'm starting this journey to build a basic MicroSaaS in public, and I was thinking - perhaps others would want to see it unfolds.

I'm just getting started without much experience coding so you can join me even if you never coded before.

So take the leap and let's see what you end up building yourself 👷🏾

Improve your product notifications with Digest, Separate Feeds, and Notification Action Buttons

Omer Ben Ami — Wed, 27 Jul 2022 14:03:00 +0000

Product Notification - What are those?

Product Notifications are everywhere - Emails, SMSs, Push notifications, Direct providers such as Slack , and In-app(known as Notification Center). Basically, product notifications are any way through which products communicate key messages to end-users.

Scale - when things get complicated

When companies grow, tens or hundreds of emails and other notifications are sent - one-time password emails, weekly digests, product updates, product-led growth (PLG) emails, member invites, and so on. Once you have more than three types of emails to manage, another layer with the complexities of managing multiple notifications templates comes up. Furthermore, as companies grow, you'll have users in different channels, all requiring custom notifications daily.

Manage notifications complexity and create a better user experience

Digest

Imagine your company has an app where users interact with each other's content. Whenever you post content, users get a notification for each and every like or comment – how tedious would that be? No user will keep your app's notifications on if that's the experience they'll get.

Digest (sometimes called "batches") is a way to aggregate individual notifications into a "super" notification at a predefined time. You can also define parameters that determine the intervals in which users get notifications. So notifications are “digested” or “batched” into one message and sent at the end of the day or the interval you define.

Call to action buttons

When you get a friend request over Facebook, you can usually see that the notification has a call-to-action inside.
Confirm or Cancel - This is the simplest form of two-way communication, which is essential for any product to have.

With Novu, you can customize those buttons, and to have as many as needed so your end-users are called to action.

Different feeds of notifications

Notification Feeds are the tabs-like feeds seen in the notification center for different kinds of notifications.
It can be:
One for likes, one for comments (like Reddit).
One for Inbox, one for following (like Notion)
One for all, One for unread (like Facebook)

What happens next?

Companies add notifications incrementally as they grow, and notifications management becomes a nightmare. Another possibility is that companies choose to build a notifications product in-house from scratch. As I see it, neither option makes sense; 99% of startups die, and building a notifications product from scratch gets you closer to becoming part of this statistic.

Novu is the first open-source notifications infrastructure built by developers for developers. It's a one-stop shop for all things notifications, just like Stripe for payments.

Developers worldwide can quickly implement our infrastructure without the hassle of building a notifications system from scratch so they can focus on what matters to their company.
https://github.com/novuhq/novu

⭐️⭐️ I am happy to announce Novu new version 0.6.0!!! ⭐️⭐️

Now you can:

Create your own user journey using the drag-and-drop workflow editor.
Put notifications into digest (batch).
Create different feeds of notifications.
Add buttons (action items) inside your notifications.

I want to thank all our amazing contributors, we couldn't do it without you!
See the full list of contributors here:
https://github.com/novuhq/novu/releases/tag/v0.6.0

Novu is looking for new contributors!

Come help us out to build the best open-source notification infrastructure, get recognized by the community, and become a Community Hero here:
https://novu.co/contributors/

We are always looking for your inputs so we can build together the best notifications user experience. Any new ideas? Write me in the comments below 👇👇👇👇