Building production AI systems that turn unstructured documents into structured, reliable data.
Based in Kathmandu 🇳🇵 · M.Sc. in Computer Systems & Knowledge Engineering · Aspiring AI researcher.
- 🔭 I work on production document-intelligence pipelines — OCR/IDP, lease & financial document processing, and RAG systems.
- 🧠 Focused on GenAI, Retrieval-Augmented Generation, agents, and fine-tuning (LoRA / QLoRA).
- ☁️ AWS Machine Learning – Specialty certified; building on Vertex AI / Gemini, AWS Bedrock, and Qdrant.
- 🎓 M.Sc. student at Pulchowk Campus (IOE, Tribhuvan University) — researching XAI, LLMs & low-resource (Nepali) NLP.
- 👨🏫 Teaching Assistant for an AI Fellowship program.
- 🌱 Currently going deeper into agentic RAG, eval harnesses, and Nepali NLP tokenization.
- 💬 Ask me about RAG architectures, FastAPI in production, IDP pipelines, or ML system design.
GenAI & ML: LangChain · Hugging Face · Vertex AI / Gemini · AWS Bedrock · LoRA / QLoRA · DSPy Retrieval & Data: Qdrant · OpenSearch · PyMuPDF · OCR / IDP Backend: FastAPI · async SQLAlchemy · PostgreSQL · REST APIs
| Area | What I'm building |
|---|---|
| Document Intelligence | Production OCR/IDP pipelines for lease & financial documents |
| RAG Systems | Serverless & FastAPI-based RAG with eval harnesses and multi-provider LLM fallback |
| Agentic AI | Agent workflows for retrieval, extraction, and reasoning |
| Fine-tuning | LoRA / QLoRA adaptation for domain-specific tasks |
"Let's turn data into reliable, production-grade intelligence."