I build LLM applications — retrieval-augmented generation, agentic workflows, and production AI systems — backed by 10+ years of experience keeping production systems reliable in enterprise IT. That combination is the throughline of my work: I don't just prototype AI features, I think about what breaks them in production and how to debug it.
Currently deepening this through Ironhack's AI Engineering Bootcamp, where I built streamTube. Since August 2025 I've also been building and shipping Mnemova, a live AI-assisted learning app, as founder and sole developer.
🔭 Featured projects
streamTube — Multimodal RAG
video assistant. Answers natural-language questions about any YouTube video by combining
timestamped audio transcription, vision-based frame captioning, and OCR — every answer is
grounded with clickable timestamp citations, and it explicitly says "not covered" instead
of hallucinating when the video doesn't have the answer.
Python · FastAPI · LangChain · LangGraph · ChromaDB · faster-whisper · React/TypeScript
Mnemova — AI-assisted spaced-repetition learning platform I
founded and built. Converts raw notes into flashcards and other exercise types using a
deterministic SM-2 scheduling engine plus an AI generation layer, with an offline-first
data layer and bidirectional Supabase sync across web and mobile.
TypeScript · Next.js · Supabase
🛠️ Technical toolkit
Python TypeScript JavaScript SQL — LangChain LangGraph OpenAI API
Anthropic API Hugging Face Transformers PyTorch — ChromaDB Vector Search & Embeddings RAG Pipelines — FastAPI React Next.js Supabase — Git Linux
Docker
🧭 Background
10+ years in enterprise IT — root-cause analysis, API troubleshooting, ITIL-aligned incident management, production-system reliability — most recently at Infosys Germany. That operational discipline is what I bring to building AI systems that hold up outside a demo.
📫 Reach me