I'm an AI Engineer with 3+ years of software development experience, including full-stack product development and applied AI systems.
My work focuses on building practical LLM-enabled applications, RAG systems, and AI-powered product features, with particular attention to retrieval quality, evaluation, testing, reliability, and production-oriented engineering.
I started from full-stack software engineering and moved deeper into applied AI. That background still shapes how I approach AI systems: not only as model integrations, but as software that needs to be measurable, testable, observable, and maintainable.
A production-oriented RAG system built with Python, FastAPI, Pydantic, Sentence Transformers, ChromaDB, Gemini, Docker, and GitHub Actions.
The project includes retrieval evaluation, deterministic citations, persistent vector storage, structured logging, request tracing, local deterministic testing, and optional bounded live LLM evaluation.
I created an 11-case golden evaluation set and used it to diagnose a retrieval failure caused by character-based chunks splitting evidence across semantic boundaries.
I then ran a controlled single-variable experiment:
Fixed character chunking → boundary-aware chunking
while keeping the embedding model, top_k=4, corpus, and evaluation dataset unchanged.
| Metric | Baseline | After change |
|---|---|---|
| Hit Rate@4 | 85.7% | 100% |
| MRR | 0.857 | 1.000 |
These results are from a small controlled evaluation set, not a claim of 100% overall RAG accuracy.
The engineering process was:
measure → inspect failure → form hypothesis → change one variable → re-evaluate
- Retrieval-Augmented Generation (RAG)
- LLM-enabled applications
- Retrieval and response evaluation
- Grounded generation and deterministic citations
- AI system testing and failure analysis
- Agentic and multi-agent workflows
- API and backend engineering for AI applications
- Production-oriented AI engineering
AI & Backend: Python · FastAPI · Pydantic · Gemini APIs · OpenAI-compatible APIs · Sentence Transformers · ChromaDB
Software Engineering: TypeScript · JavaScript · React · React Native · Node.js · REST APIs
Engineering & Delivery: Docker · GitHub Actions · Automated Testing · Structured Logging · Request Tracing
I'm continuing to go deeper into Applied AI, LLM systems, RAG evaluation, and agentic AI, while building on my full-stack software engineering background.
I'm particularly interested in engineering AI systems that are not only functional, but also measurable, reliable, testable, and useful in real products.