https://neuralbroker.github.io/ Notes on fine-tuning, production ML, RAG, and system design. en Sat, 28 Jun 2026 00:00:00 +0000 https://neuralbroker.github.io/writing/projects-and-learnings.html https://neuralbroker.github.io/writing/projects-and-learnings.html Wed, 04 Jun 2026 00:00:00 +0000 Walkthrough of BlitzKode, EsenceLab, Lexora AI, and SentinelML — design decisions and what I'd do differently. https://neuralbroker.github.io/writing/fine-tuning-is-mostly-data-work.html https://neuralbroker.github.io/writing/fine-tuning-is-mostly-data-work.html Sat, 30 May 2026 00:00:00 +0000 Lessons from building BlitzKode — why SFT data quality matters more than model architecture. https://neuralbroker.github.io/writing/your-ml-system-will-fail-in-production.html https://neuralbroker.github.io/writing/your-ml-system-will-fail-in-production.html Sun, 25 May 2026 00:00:00 +0000 SentinelML postmortem — drift detection, retraining, and why monitoring beats model cards. https://neuralbroker.github.io/writing/reading-note-neurosymbolic-ai.html https://neuralbroker.github.io/writing/reading-note-neurosymbolic-ai.html Tue, 11 Feb 2025 00:00:00 +0000 Reading notes from a seminar paper on neurosymbolic AI approaches.