I am the one, who can make your AI better.
I research optimization, orchestration, environment, memory and context use in AI agents and multi-agent systems and lead teams to build impressive products for Fortune 500 companies and startups.
- 🔬 Applied Research Intern @ JetBrains Research — studying context engineering for AI coding agents (Claude Code, Cursor, Copilot). Large-scale repo mining, embedding-based clustering, LLM-assisted pattern extraction
- 🏗️ Data & AI Software Engineer @ KI performance — shipped multi-agent LLM systems for Microsoft, Lufthansa, Mercedes-Benz, DHL. LangGraph orchestration on Azure. Certified: Azure AI Engineer, Azure Developer, Databricks Developer
- 📄 Published a systematic literature review on agentic systems optimization — five-layer taxonomy across 32 primary studies
- 🚀 Founded PUM — 50+ member builder collective, 15+ international hackathons, 10+ shipped prototypes under 24–48h constraints
- 🎓 B.Sc. Computer Science @ TUM — 5th semester, 149 ECTS (125% of recommended pace).
- Multi-agent system optimization — architecture, evaluation, performance
- Context & environment engineering for AI coding agents — configs, memory, orchestration
- Prompt analysis & refinement — pattern recognition, quality metrics, debugging
15+ international hackathons · 10+ shipped systems · 50+ team members
Built systems across:
- multi-agent RAG platforms
- real-time data + ML pipelines
- accessibility & navigation
- developer productivity tools
Won at hackathons, datathons and ideathons with international teams as teamlead, main ML engineer and solution architect. Founded a hackathon organization.
👉 Full portfolio: https://devpost.com/smrtkozyrev
I enjoy high-intensity environments where good systems thinking matters — especially hackathons, research-driven prototyping, and interdisciplinary research-heavy builds that need to go from idea to demo fast.