The Engineering Notebook — page 3 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.
Developer Tools LM Studio Power User Setup 2026: Profiles, Quants, LAN Serving
My opinionated 2026 LM Studio setup: reproducible model profiles, quant hygiene rules, scripted loads via lms + /api/v1, and a secure multi-user LAN server.
Cybersecurity How to Ship EU AI Act Compliance for AI Agents (Minimal Pack)
A shippable, vendor-neutral evidence pack for AI agents: run logs, tool approvals, guardrail configs, retention/redaction, and an audit export you can produce in days.
Developer Tools MCP vs Function Calling in Agents [2026]: When to Say No
MCP backlash is real. Here’s my decision framework for when to use MCP vs function calling in agents, and when a bespoke tool API beats both on ops, security, and debugging.
AI and Machine Learning How to Do Synthetic Data for RAG Evaluation [2026 Pipeline]
A leakage-first, adversarial-negative pipeline to generate, label, and validate a synthetic RAG evaluation set without building a benchmark that only tests your generator.
Cybersecurity How to Add AI Agent Kill Switch Spend Limits [2026]
A software-only watchdog pattern for AI agents: a single enforcement point for kill switches, per-tool budgets, approval gates, and tamper-evident audit logs.
Cloud and DevOps vLLM Self-Hosted LLM Production Checklist [2026]: Auth + Quotas
A production-first checklist for self-hosting an OpenAI-compatible vLLM endpoint: auth, per-tenant quotas, streaming SSE, queueing, and redaction-safe logs.
Technology How to Run Unsloth on AMD GPUs (ROCm) [2026 Guide]
A late‑2026 compatibility matrix plus known‑good commands for getting Unsloth LoRA fine‑tuning to actually converge on Radeon and MI GPUs with ROCm.
Technology Intel Arc B‑Series Local LLM Benchmark [2026]: Arc Done Right
A reproducible Intel Arc B‑series local LLM benchmark harness (TTFT, tok/s, VRAM, power), plus a Linux/Windows compatibility matrix and Arc-friendly GGUF quants that won’t OOM at 8K context.
![ollama terminal macbook laptop screen — illustration for article on Ollaya Ollama Decision Model Setup [2026]:](https://img.kunalganglani.com/images/vzekdneq/production/a6ede8c9d4b1546ce5b4e28f601f2cf78d2b4815-1200x675.webp?auto=format&fit=max&q=75&w=600)
Developer Tools Ollaya Ollama Decision Model Setup [2026]: Routing + Benchmarks
Set up Ollaya locally, write safe routing questions, point the TypeSafe SDK at localhost, and benchmark p50/p95 (warm + cold) vs LLM JSON routing.
Developer Tools AI Generated Code Quality [2026]: A 0–100 Audit Rubric + CI Gates
AI-generated code quality isn’t a style problem anymore. Here’s a practical 0–100 audit rubric (tests, security, maintainability, rewrite risk) plus CI gates that block hallucinated APIs before they ship.
Cybersecurity MCP OAuth Security: Tool Impersonation, aud Mismatch, Token Replay [2026]
A threat model for MCP tool servers using OAuth: how tool impersonation and audience mismatch happen, how tokens get replayed, and what to validate and log in 2026.
Frontend and Mobile How to Build a Browser Based AI Agent (Safe-by-Default) [2026]
A practical browser AI agent architecture that stays local: Worker planner, Wasm sandbox, IndexedDB memory, and a service worker tool bridge with real guardrails.