LLM applications
Assistants, document Q&A, and domain workflows shaped around real user behavior and guarded by evaluation.

Applied AI Engineer · Kathmandu
I build practical AI systems — RAG pipelines, agent workflows, and production APIs — for teams that need reliable results instead of flashy demos.

What I build
The work is strongest when product thinking and systems engineering meet: fewer demos, more usable decisions.
Assistants, document Q&A, and domain workflows shaped around real user behavior and guarded by evaluation.
Parsing, semantic chunking, hybrid retrieval, reranking, and citations that make answers inspectable.
LangGraph orchestration, tool selection, memory, and streaming states for tasks that need more than one step.
FastAPI backends, async pipelines, containerized deployments, and feedback loops after release.
Selected work
A few builds that combine AI behavior, backend reliability, and a practical user workflow.
Multi-agent research pipeline that produces citation-grounded research briefs with real-time streaming.
RAG-powered code review agent that provides context-aware suggestions and automates PR reviews via GitHub Actions.
Capabilities
Filter by discipline to see what I use on real projects.
30 tools
Whether you're looking for an AI engineer, want to collaborate on research, or have an interesting problem, I'd love to hear from you.