A public notebook. Not random blog posts. A canon in progress.
Agents, cognitive debt, and the future of governed work. The central thesis.
From Q&A chatbots to systems that independently plan, retrieve, act, and verify. Latency, cost, quality tradeoffs at each stage.
Transparency, control, and building agents that search with intention rather than brute force.
NBA stats, Balatro, and the cost of curiosity. Eight iterations of benchmark design, from kitchen sink to meaningful model separation.
How Qdrant's native MMR balances similar results with diverse results for better retrieval quality.
Breaking echo chambers in search results using the DeepFashion dataset as a proving ground.
Turning scanned documents into a searchable knowledge base with OCR + vector indexing.
Reproducible RAG evaluation using Qdrant for retrieval and Relari for real-world performance assessment.