A fast, hand-built portfolio of shipped RAG and multi-agent AI systems — every capability backed by a real project and honest metrics.
Four capability areas, each proven by shipped work:
- RAG Systems — chunking, embeddings, vector retrieval, page-level citations, and refusal when the answer isn't in the documents
- Multi-Agent Systems — tool loops, structured output with validation, provider fallback (hand-rolled, no framework)
- Evaluation & Metering — LLM-as-judge eval suites and per-request cost metering
- Deployment & Infra — containerized and shipped
| Project | What it is |
|---|---|
| CareerAgent | Multi-agent platform for the job hunt |
| Multi-agent root-cause analysis | Master thesis @ Siemens (industrial AI) |
| RetrievalLab | Advanced RAG, made visible |
| ai-rag-project | RAG — chat with your documents |
React · TypeScript · Vite — a single-page site with client-side routing and a small markdown blog.
npm install
npm run devThen open the local URL Vite prints (e.g. http://localhost:5173).