I build web applications end to end — Node.js and Express on the server, React and Next.js on the client — with a foundation in data structures and algorithms from C and C++. I have been shipping for the web since 2022 and currently work as an SDE at Compro Technologies.

Lately my focus has shifted toward applied AI engineering: retrieval-augmented generation, vector search, and the queue-driven pipelines that make document ingestion work at scale. I care about systems that fail honestly — background workers that can crash and restart, answers that cite their sources instead of hallucinating, and configuration that does not hide surprises.

• 🌍 Based in New Delhi, India
• 💼 Senior SDE at Compro Technologies
• 🧠 Currently going deeper on RAG pipelines, vector databases, and LLM orchestration
• ✉️ Reach me at singhavnish516@gmail.com
• 🔗 Resume

PDF RAG — upload a PDF, ask questions about it, get answers grounded in the document with page-level citations.

A full retrieval-augmented generation pipeline rather than a wrapper around a chat API. A PDF is parsed, chunked, and embedded into a vector store; each question retrieves only the passages that matter and hands them to the model as context, so answers cite real pages. Ingestion runs through a job queue on a separate worker process, so a slow document never blocks the HTTP request.

_{Next.js 15 · React 19 · TypeScript · Tailwind 4 · Express · BullMQ on Valkey · Qdrant · LangChain.js · ONNX embeddings · Ollama · Docker Compose}

Languages

Frontend

Backend & data

AI & tooling

_{Also working with: LangChain.js · Qdrant · BullMQ · Valkey/Redis · Ollama · Clerk}