Hi, I'm Nirbhay Singh
AI/ML Engineer building production-grade GenAI systems — custom LLMs, RAG pipelines, and real-time NLP agents used by real teams
Engineer obsessed with shipping intelligent systems
I design and build production-grade AI applications — from retrieval pipelines and LLM orchestration to the APIs and interfaces that bring them to users.
I'm a software engineer focused on the GenAI stack: LLM architecture, RAG pipelines, sentiment analysis systems, and the full-stack plumbing that makes them production-ready.
I love taking research-grade ideas into production. I trained Rudra, a custom 6.5M-parameter Transformer LLM with autonomous tool-use, on a single consumer GPU. At TATA BlueScope Steel, I built a FinBERT-powered sentiment pipeline processing 500+ financial articles daily at ~82% accuracy — cutting manual analyst effort by 60%.
What I focus on
A timeline of shipping work
Roles where I've built production AI and full-stack systems.
Software Engineer
Anantron International Pvt Ltd
- Architected production REST APIs serving real workloads
- Stack: Node.js, Express.js, FastAPI, React.js
- Owned performance optimization across hot paths
- Shipped production deployments end-to-end
AI / ML Intern
TATA BlueScope Steel
- Built a FinBERT-based NLP pipeline for financial news
- Processed 500+ articles per day with 82% sentiment accuracy
- Delivered Streamlit dashboards for stakeholders
- Automated end-to-end ingestion → inference workflows
Production AI work, shipped
Selected systems combining LLMs, retrieval, and well-engineered backends.
Project 1
Rudra — Custom AI Agent & MiniGPT
Custom 6.5M-parameter Transformer LLM with a ReAct agent framework for autonomous tool use, trained from scratch and deployed for real-time inference.
Project 2
AI Stock Sentiment Analyzer
Real-time financial news pipeline with FinBERT scoring and LLM-generated executive summaries.
Project 3
Movie Recommendation System
Hybrid recommender over 45k+ titles with TF-IDF vectorization and cosine similarity.
Project 4
AI PDF Chatbot
Production RAG pipeline with semantic search, multi-turn memory, and sub-2s responses.
The stack I build with
Practical, production-tested tools across the AI and full-stack spectrum.
Continuous learning, certified
Oracle Generative AI Professional
Oracle
OCI Foundations Associate
Oracle
HuggingFace LLM Fundamentals
HuggingFace
Google Cloud Generative AI
Google Cloud
TensorFlow Deep Learning Fundamentals
TensorFlow
IBM LLM Fundamentals
IBM
IBM NLP & Computer Vision
IBM
JP Morgan Quantitative Research
JP Morgan
Let's build something
Open to roles, collaborations, and consulting on AI products.
Have an idea?
Let's make it real.
Whether it's a startup, a side project, or a full-time role — I'd love to hear what you're building.
Usually replies within 24 hours