Building AI Systems
Bangalore, India
I am looking to join a strong team in a role related to AI. My preference is a fast moving, small team, somewhere between 1 and 100 people, building something people actually want, not an idea born out of thin air just for the sake of doing a startup. I can join immediately. You can reach me at nikm5502@gmail.com.
I love tinkering and keep building things on the side, sometimes just to learn, sometimes to see how the technology is moving.
Frankly, if you spend time on my portfolio, you will see broad experience across the stack, and I know that costs depth in any one domain. I realize that too, and I want to go deep in one area.
I worked across the full stack for a long time, including production deployments on AWS and building pipelines in my previous roles. I also spent time on diffusion and fine tuning, including writing architectures in PyTorch. Since then I have switched gears to focus fully on building AI systems. This is not about the hype. It is fun to watch AI become an enabler and scale both horizontally and vertically across products, and my mix of full stack work with an architecture level understanding of AI turns out to be useful for building systems around it. That feels like my edge.
Right now I am building a few layers from scratch to strip away the abstractions these frameworks hide. Turns out simple patterns win.
I also have experience working in fast paced cultures, across engineering and product, leading small teams and building products end to end, from shaping the idea to writing the code to shipping and improving it.
A coding agent built with no framework: ReAct loop, custom tool layer, context compaction and a Docker runtime. Resolved 26 of a 30 instance SWE-bench Verified sample, with tracing and LLM as a judge evals to catch trajectory regressions.
A key value store written in C. Concurrent TCP server, RESP parsing, key expiry, and master replica synchronization.
Trained a VAE from scratch to generate and interpolate hand drawn sketches, then wrote up how the training and inference actually work.
Built the dataset by scraping sites, running OCR on scanned books and transcribing YouTube audio, generated question answer pairs with the Gemini API, then fine tuned Gemma-3-1B-IT on rented GPUs.
A LangGraph agent that asks clarifying questions first, then scrapes Amazon, Flipkart and Google Search to compare products and return the top three matches.
The official platform for campus elections at IIT Roorkee, where 10K+ students voted inside a two hour window. Tamper resistant voting, and the AWS infrastructure that held through the peak.
Built for the Indian Knowledge Systems division at the Ministry of Education. Upload a certificate template, point it at a spreadsheet, and it generates personalised certificates and emails them out, with SQS and Lambda doing the work in the background.
Led technical projects end to end, and ran workshops, lectures and mentoring for 100+ students across the club.
A runnable script walked through line by line to show what distributed training actually does: synchronization, seeding, and how the GPUs talk to each other.
A step by step breakdown of the inference pipeline, from the T5 encoder through the latent DiT to the VAE decoder.
V8, libuv and the event loop, and why the event driven architecture is built the way it is.
Every file is the same binary digits. What decides whether they are an image, audio or text is only how you agree to read them.