// portfolio + research log
Ut_Singh
Backend engineer building Agentic AI systems.
Plan — decompose the goal into steps
// currently learning
- Retrieval-augmented generation (RAG) pipelines
- LangGraph agent orchestration & control loops
- Evaluation harnesses for LLM systems
- Vector databases & hybrid search
Deep-dives land in the Research Log. What I am on right now lives on the Now page.
// recent notes
- LangGraph agent loopsModeling an agent as a graph with cycles: how LangGraph turns the plan → act → observe → reflect loop into explicit, inspectable state transitions.
- RAG retrieval basicsThe retrieval half of retrieval-augmented generation — chunking, embeddings, similarity search, and why hybrid retrieval usually beats pure vectors.