ML engineer in London building AI systems that (mostly) do what they're told.

I work at the intersection of AI, Agentic systems, and production-grade enterprise systems, mostly wrestling LLMs into doing useful things inside production environments where the stakes are higher than a weekend hackathon.

Right now that means:

• Building Graph RAG agents with hybrid BM25 + vector search, Azure AI Foundry, and a healthy fear of hallucinations
• Designing document intelligence systems that process legal and patent documents at scale
• Exploring LLM evaluation strategies because vibes-based QA doesn't scale
• Occasionally questioning every architectural decision at 11 pm

• 🎓 Graduated from Queen Mary University of London
• 🏙️ Side-questing on different AI tools
• 📚 Obsessive about personal knowledge management and the tools around it
• 🤔 Convinced that most software problems are, at their core, a data modelling problem in disguise

If you're into RAGs, LLMs, NLP, or want to hear why my chunking strategy is better than yours (it isn't), find me on Telegram or drop by sepehr.uk.