$ whoami
Vishwajit Avarsekar — backend-leaning full-stack developer with interest in AI.
$ cat status.txt
$ Backend-leaning engineer, shipped a fine-tuned Nutri-Score classifier, a RAG system, an npm package, and a full-stack platform. Looking to join an early-stage team.
$ ls interests/
backend-systems/ databases/ api-design/ applied-ai/ rag-pipelines/
$
I'm Vishwajit Avarsekar, a backend-leaning full-stack developer. I have written production code at an early-stage startup. Since then I've been building independently.
I've built: a real estate platform end to end on the MERN stack, an npm package that profiles where API latency actually goes, freelance work delivering a listing site for a client, and — this week — a retrieval-augmented system(RAG) that answers questions over Indian food-safety regulations.
My background is backend: APIs, databases, the parts of a system that aren't visible in a demo but decide whether it stays up. Lately that's extended into applied AI — retrieval, embeddings, and wiring LLMs into products that need to be correct, not just impressive.
I'm open for opportunities preferably at an early-stage environment, I believe in code ownership with real stakes. I'm currently in Bengaluru, India. but open to remote work.
A fine-tuned DistilBERT classifier that predicts a food product's Nutri-Score grade (A–E) from ingredient text alone, with labels self-computed from the official formula rather than trusted from source data. Trained on ~926k global products, evaluated separately on a held-out India set to surface a real domain-shift gap.
A RAG system over Indian food-safety (FSSAI) regulations — ask a question in plain language, get an answer grounded in the actual regulatory text. Uses hybrid search (BM25 + vector embeddings) so retrieval doesn't miss exact clause matches the way pure semantic search can.
Published npm package that breaks down the total latency of a request or query into DB latency, server/compute latency, and network overhead — so you know where the time actually went instead of guessing. More instrumentation in progress.
Full-stack real estate listing platform with search and multi-criteria filters, built solo end to end — data model, REST API, auth, and the React frontend.
Freelance engagement building a listing website for a client — worked requirements directly with the client and delivered iteratively. Code is client-owned, so no public repo, happy to walk through it live.
A few smaller things — basic Web Crawler, Imitating Birds flying pattern with Java, Weather forecaster — live on GitHub.
Delivered a listing website for a client, working requirements directly with them and shipping iteratively rather than against a fixed spec. maintaining current infrastructure with new features in the pipeline. Code is client-owned, so no public repo, happy to walk through it live.
Worked across the stack on a production product, leaning backend — APIs, database design, and the services behind the parts users actually see.
Node.js · Express · REST APIs · Go (basic) · fast API · Spring Boot
MongoDB · PostgreSQL · OpenSearch · Vector search
RAG pipelines · Embeddings · Hybrid search
React.js · JavaScript · HTML/CSS · Next.js
Git · Docker · Linux · npm publishing
Open to full-stack and applied-AI roles at early-stage teams. Reach out at vishsecures@gmail.com, or find me here: