DEV Community: Somay

Building RabbitHole (Even While It Breaks)

Somay — Fri, 19 Jun 2026 16:26:47 +0000

Some projects reach a stage where they're not "finished", not even close, but they're finally alive.

RabbitHole reached that point today.

There are still more bugs than I'd like to admit, conversations occasionally go off the rails but the core infrastructure is finally in place, and that's worth celebrating.

Current progress

Multi-agent architecture with LangGraph
Perspective generation and courtroom-style debates
State management and memory layers
Human-in-the-loop flow
Graph orchestration running locally

There are still countless things left to improve, but the foundation exists now.

Another thing that made this month special:

I was fortunate enough to make it to the second round of two different fellowship programs. Regardless of the outcome, I'm grateful that something in my work and ideas resonated with people.

One thing I've been learning recently:

You don't need to wait until a project is perfect before sharing it.

Sometimes progress deserves to be documented while things are still messy.

A lot of engineering is just:

Build.
Break.
Debug.
Repeat.

And honestly, that's where most of the fun is.

RabbitHole Repository

🔗 GitHub: https://github.com/Somay-kousis/RabbitHole

Still a long way to go, but we're moving.

RabbitHoles while learning AI

Somay — Sun, 14 Jun 2026 07:46:42 +0000

I was supposed to stop at LangGraph.

That was the plan.

Instead, the rabbit hole decided otherwise.

Some of the things I unexpectedly picked up while chasing Agentic AI:

⚙️ Docker

Multi-stage Dockerfiles
Docker Compose
Volumes
Building images from my own projects instead of just pulling them
Finally understanding why "works on my machine" isn't enough

🤖 GitHub Actions

Automated builds and tests
CI/CD pipelines
Branch protection rules
Secrets management
Automatic Docker image publishing
Realizing how satisfying it is when pushing code triggers everything for you

🌿 Git & Engineering Practices

Conventional commits
Pull requests even for solo projects
Trunk Based Development
Small, rapid merges instead of week-long branches
Keeping main always deployable

🚀 Deployment

Actually shipping things instead of leaving them in localhost prison

And somehow, while trying to learn AI, I ended up appreciating software engineering just as much.

Still left:

• LlamaIndex
• Cloud
• Terraform

Apparently the roadmap had other plans.

Curious what unexpected rabbit holes everyone else fell into while learning AI.

Built my first proper agentic AI project

Somay — Wed, 10 Jun 2026 09:39:11 +0000

Over the last few weeks, while learning LangGraph and agentic systems, I ended up building Co-Founder Memory.

It's a stateful AI assistant with:

• long-term memory

• planning loops

• self-correcting RAG

• web search fallback

• automated timeline summaries

• project and preference tracking

Nothing revolutionary — many ideas already exist. The goal wasn't to reinvent memory, but to understand how these systems work by actually building one.

A lot of concepts only started making sense once I had to connect them together:

graph-based workflows with LangGraph
memory extraction and storage
retrieval and validation loops
routing and planning nodes
maintaining context across sessions

Building it taught me far more than watching tutorials ever did.

Repo:
https://github.com/Somay-kousis/Co-Founder-Memory

I'm currently entering my 3rd year at IIITM Gwalior and looking for ML / GenAI internships. If you're building interesting things around LLMs, agents, RAG, or AI products, I'd love to connect.

Always happy to chat with fellow builders as well 🚀

AI #GenerativeAI #LangGraph #RAG #LLM #MachineLearning #Internship

A few months ago, I wouldn't have picked myself

Somay — Tue, 02 Jun 2026 03:24:06 +0000

Back in February, a friend asked me to join his hackathon team.

My first reaction wasn't excitement.

It was:

"Can I even contribute anything?"

I remember repeatedly telling him not to add dead weight to the team and to find someone better. He kept insisting that it didn't matter and that I should just join.

The funny thing is, I still don't think I've done anything extraordinary since then.

No big startup.
No crazy achievement.
No overnight success story.

Mostly just hundreds of hours of learning, building random things, breaking them, fixing them, and realizing how much I still don't know.

But today I caught myself doing something weird.

I'm the one thinking about who to bring into a team.

And for the first time, I don't immediately feel like I'd be dead weight.

Not because I know everything now.

Just because I've reached the point where I can look at a problem and genuinely believe that, given enough time, I'll figure out how to contribute.

It's a small shift, but it feels important.

A few months ago I was wondering if I belonged on a team at all.

Today I'm wondering who should be on mine. 👀

Can you guys lemme know which of the following tagline is better 👀

Somay — Mon, 01 Jun 2026 14:34:30 +0000

Pick it for a platform named “something” where people build stuff

A) Start with something.
B) Something is about to happen

( I didn't share details cause you'll see it once it is launched )

Today My Portfolio Started Talking Back 👀

Somay — Sat, 30 May 2026 10:58:54 +0000

Today was one of those weird days.

the AI portfolio bot is technically alive now 😭

it can talk about my projects, writing, ambitions, weird philosophies, random life decisions, and somehow even picked up on things i never explicitly taught it.

repo:
https://github.com/Somay-kousis/self.so

portfolio:
https://portfolio-sable-psi-56.vercel.app

the funny part is that building the actual bot wasn't the hardest thing.

the hardest thing was finding free stuff.

free embeddings.
free vector db.
free llm.
free hosting.

every tutorial casually says "just use xyz api" and then you discover xyz wants your kidneys after the trial ends 💀

so half of the project was AI engineering and the other half was surviving the free tier ecosystem.

also ran out of Groq tokens today.

which feels poetic because the bot became smart enough to have conversations right when its brain got disconnected 😭

meanwhile:

slight headache for days
CG ended up lower than expected
agentic AI starts next
internship season is getting closer
still have DSA waiting for me in a dark alley

but okay.

one thing at a time.

a few weeks ago i was learning what document loaders even were.

today i deployed something that can talk about me better than i can talk about myself on some days.

that's kinda cool.

I think AI accidentally became my personality for a month

Somay — Wed, 27 May 2026 07:48:21 +0000

Okay so I think I’m finally done with the GenAI foundations phase

like genuinely done

RAG, tools, tool calling, agents, Ollama and all that stuff that sounded terrifying 2 weeks ago now somehow makes sense in my head

and the weird thing is… I don’t think AI feels like “chatbots” anymore

it feels more like systems now

like workflows
memory
humans
context
behaviour
retrieval
decision making
all stitched together somehow

which is lowkey why I got obsessed with it in the first place

also had a very funny realization recently

I don’t think I’m the kind of person who wants to spend all day training models and doing hardcore DL research

I like the human side of AI wayyyyy more

the part where:
an AI can understand someone’s tone,
or remember context,
or connect similar people,
or help communities form,
or act differently depending on emotions/situations

basically I like AI when it starts feeling less like software and more like a system interacting with humans

which probably explains why my project ideas always become psychological somehow

anywayyy

next phase starts in June:
Agentic AI + LangGraph + MCP

and I’ll be building 2 projects alongside:

a portfolio AI assistant
Signal (still evolving in my head but it’s about people, intent, compatibility and systems)

also somewhere in between all this:
DSA
internships
open source maybe
foreign applications maybe
and whatever other personality crisis I decide to develop 👍

skills wise rn:
GenAI
RAG
LangChain
LangGraph (starting)
Agents
FastAPI (soon)
MCP (soon)
React
Next.js
TypeScript
Tailwind
UI/UX
Product thinking
presentations
design
writing
community building
and apparently overthinking

if anyone knows internships / people building cool AI stuff / open source routes I should explore, pls let me know 👀

Looking for 4 people to build something weird with me

Somay — Mon, 25 May 2026 14:52:53 +0000

A few months ago, if someone had asked me what I'd be doing this summer, I would've probably said DSA.

104 LeetCode problems later, I somehow ended up obsessed with AI, simulations, product design, human behavior, and all the strange ways technology can teach people things.

Recently I came across the IEEE Metaverse Grand Challenge for Simulation-Based Learning.

The premise is simple:

Build an immersive simulation that teaches something.

Not another chatbot.
Not another CRUD app.
Not another dashboard pretending to be a startup.

A simulation.

Something people can step into, interact with, and learn from.

I don't know exactly what we're building yet.

Maybe a public safety simulation.
Maybe a smart city that reacts to your decisions.
Maybe a healthcare training environment.
Maybe something completely different.

What I do know is that I don't want to do it alone.

So I'm looking for 4 people who want to spend the next few months building something ambitious.

Current stack on my side:

React / Next.js
Product Design
UI/UX
Presentations & Pitching
Generative AI
FastAPI (currently learning)
Agentic Systems (currently learning)

People I'd love to meet:

Unity developers
Blender artists
AI / ML engineers
Backend developers
Anyone obsessed with learning experiences, simulations, psychology, education, or human behavior

Experience matters less than curiosity.

The deadline is August 15.

If we end up building something cool, great.

If we end up learning a ridiculous amount and making new friends, that's probably a win too.

If this sounds interesting, reach out.

Let's build something weird.

somaykaush@gmail.com (mail me if interested or just comment here)

Building: Women's Safety Isn't About Women

Somay — Sun, 24 May 2026 11:34:54 +0000

I'm starting work on a project I've been thinking about for a long time.

The tagline is:

"Women's safety isn't about women."

At first glance, that sounds strange.

But the more I looked at the problem, the more I felt that most solutions focus on what happens after risk appears:

Emergency buttons
Location sharing
SOS alerts
Reporting systems

These are important.

But I became interested in a different question:

What if safety is influenced more by the people, communities, and environments around someone than by the tools they carry?

This project is not another social platform.

It's not a reputation system.

It's not a public scoring mechanism.

And it definitely isn't about replacing human judgment with AI.

The goal is to explore whether technology can help create accountability around harmful behavior while still respecting privacy, human oversight, and due process.

I'm still in the design and architecture phase, so I won't be sharing implementation details yet.

For now, I'm documenting the journey, the technical challenges, the ethical questions, and the lessons learned while building it.

If you're interested in privacy, AI systems, trust & safety, platform design, moderation, cybersecurity, or social impact technology, feel free to follow along.

The first step starts now.

Women's safety isn't about women.
It's about the world around them.

I Finished Machine Learning. And Then Changed The Plan.

Somay — Fri, 22 May 2026 14:51:02 +0000

Dev.to / Medium / Substack

I Finished Machine Learning. And Then Changed The Plan.

A few months ago, I had no idea what feature engineering was.

Today, I finished my Machine Learning roadmap.

Not "mastered ML."
Not "became an AI expert."

Finished the part where every tutorial starts making sense.

I built projects, broke models, overfit them, leaked data into them, fixed them, and slowly started understanding why things worked instead of blindly following notebooks.

My latest project:

Customer Churn Prediction

Predicting which customers are likely to leave a company before they actually do.

Built with:

Python
Pandas
Scikit-Learn
XGBoost
Feature Engineering
Hyperparameter Tuning

Project:
Customer Churn Prediction Demo

The funny thing?

The more I learned, the less interested I became in rushing toward Deep Learning.

Originally the plan was:

ML → Deep Learning → NLP

But somewhere along the way I realized something.

I don't just want to understand models.

I want to build products.

Things people actually use.

So the roadmap changed.

Now I'm diving into:

Generative AI
RAG
LangChain
FastAPI
Ollama
MCP
LangGraph
Agentic AI

Deep Learning isn't gone.

It's just waiting its turn.

And DSA?

That was supposed to stay consistent.

Instead, I keep finding myself opening AI documentation at 2 AM and disappearing into another rabbit hole.

Not because I have to.

Because I genuinely can't stop.

Somewhere between building projects and studying, curiosity quietly took over.

So that's where we are now.

Machine Learning: complete.

Next stop: GenAI.

Let's see how deep this rabbit hole goes.

If you're earlier in your journey:

Build projects before you feel ready.

Most of what I learned came from fixing mistakes I didn't know I was making.

Project:
Customer Churn Prediction Demo

I rebuilt my little corner of the internet

Somay — Thu, 21 May 2026 01:41:20 +0000

Spent the last day redesigning my github + portfolio until they finally started feeling like me instead of another developer template pretending to have personality.

dark interfaces.

tiny details nobody notices consciously.

machine learning mixed with interaction design.

and a slightly unhealthy obsession with atmosphere.

currently somewhere between:

frontend engineering
machine learning
and “what if software felt haunted in a good way?”

portfolio:

Somay-kousis (Blink) · GitHub

Bleh Bleh Bleh. Somay-kousis has 10 repositories available. Follow their code on GitHub.

github.com

github:

portfolio-sable-psi-56.vercel.app

still evolving, but i think it finally feels alive now.

btw guys I’m not inconsistent anymore?

Somay — Thu, 21 May 2026 01:28:55 +0000

for the past 2 weeks, life has mostly been:

wake up → laptop → confusion → somehow figure things out → repeat

and surprisingly… it’s been working.

i’ve been averaging around 8 actually productive hours almost every day

(except one break day because my brain started sounding like a dying laptop fan)

recently:

prettified my portfolio + github
finished my first ML project
started learning graphs + logistic regression today
restarted workouts because my wrist is probablyyyyy functional again

lowkey feels nice seeing things compound for once.

if things keep going like this, by next month i should be deep into:

dynamic programming
deep learning
and whatever identity crisis comes with them

still feel lost half the time btw.

just slightly more organized now.