The Engineering Notebook — page 6 of 38
Notes on building with AI, agents & the modern stack.
Deep dives on AI/ML, RAG systems, agent engineering, and senior-engineer architecture decisions.
AI and Machine Learning RAG Evaluation Metrics for Retrieval Quality: My Production Playbook
If your RAG app got worse after an embeddings or chunking change, grading answers won’t tell you why. Here’s a retrieval-first eval workflow: recall@k, MRR, citation accuracy, leakage tests, and a frozen offline corpus so regressions are real—not vibes.
Developer Tools Review AI-Generated Code Checklist [2026]: Beat the Bottleneck
AI makes code cheap. Verification is the new tax. Here’s a practical workflow, checklist, and ‘delete & redo’ rubric to keep PRs fast and safe in 2026.
AI and Machine Learning How to Set Google Workspace AI Agent Permissions + Audit Trail [2026]
A practical, least-privilege model for Drive/Gmail agents: scopes, per-user vs domain-wide delegation, app access control defaults, and an incident-ready audit trail.
Developer Tools How to Benchmark AI Coding Tools on Your Own Repo [2026]
A reproducible way to benchmark AI coding tools on your own repo using SWE-bench-style tasks, human baselines, defect scoring, and anti-gaming rules.
AI and Machine Learning Prompt Compression Token Savings Benchmark [2026]: RTK vs Reality
Token savings are easy to market and hard to bank. Here’s the benchmark framework I trust for RTK-style compression: cost-per-success, tool-call failures, and latency.
AI and Machine Learning How to Ship OpenAI Agents API Guardrails in 1 Day [2026]
Production tool-use fails on governance, not prompts. Here’s a practical setup for OpenAI Agents API guardrails: allowlists, layered rate limits, and audit logs you can actually use in incidents.
Developer Tools Bun lockfile bun.lockb format: Threat model + CI checks [2026]
bun.lockb made Bun installs fast, but it also changed how you audit dependency drift. Here’s what’s inside, what integrity it gives you, and how to enforce it in CI.
Cybersecurity Why AI Voice Detectors Fail [2026]: Codecs, Watermarks, Traps
“AI voice detector” scores swing wildly because the audio channel is the adversary. Here’s how codecs, noise suppression, and bad metrics break detection in the 2026 real world.
Developer Tools How to Secure MCP Servers: Auth + AuthZ [2026 Tutorial]
A gateway-grade blueprint for MCP servers: authenticated discovery, PKCE/OAuth done right, tool-scoped permissions, audit logs, rate limits, and secret handling—with runnable Node/Express code.
Developer Tools Rust Reproducible Builds + SBOM + Signed Artifacts [2026]
A 2026-ready Rust release pipeline: pin toolchains, make builds reproducible, gate deps with cargo-vet, generate SBOM + SLSA provenance, and sign everything with keyless cosign.
![markdown documentation code editor laptop screen — illustration for article on Agent Readable Documentation Toolchain [2026]:](https://img.kunalganglani.com/images/vzekdneq/production/5d5a338107b4aa20b1df7ae6932d19901d946eba-1200x675.webp?auto=format&fit=max&q=75&w=600)
Developer Tools Agent Readable Documentation Toolchain [2026]: My Stack
A docs-as-code pipeline that stays great for humans, useful for your trusted AI agents, and aggressively unhelpful to random scrapers. Practical, not theoretical.
Technology 12 Possible Endings for AI: Fact-Checking the “MIT” Meme [2026]
The viral “MIT Explains the 12 Possible Endings for AI” claim isn’t an MIT publication. Here’s a grounded 12-scenario taxonomy, signals to watch, and how to steer outcomes.