Applied AI Engineer with a strong full-stack background, focused on building production-grade RAG and LLM systems.

I work at the intersection of:

• Retrieval-Augmented Generation (RAG)
• Embeddings & semantic search
• Backend systems & APIs
• Product-focused AI engineering

• Designing end-to-end RAG systems for large document collections
• Building ingestion, chunking, and embedding pipelines
• Semantic & hybrid retrieval with citation-aware context assembly
• Streaming LLM responses integrated into real products
• Making RAG systems observable, tunable, and maintainable

• LLMs & RAG: OpenAI-compatible APIs, local models
• Backend: FastAPI, Node.js
• Data: PostgreSQL, vector databases (Qdrant / FAISS)
• Infra: background workers, async pipelines
• Frontend: React / Next.js (when product UX matters)

RAG is a system, not a prompt.
Data quality beats clever prompting.
Production constraints shape good AI.

• RAG Platform — production-oriented retrieval system for unstructured data
• Observer — AI-native tooling for working with complex data flows
• OSS libraries used in production products

New York, NY
https://github.com/keske