sandy15.in

Know me

Sandeep Baligeri

AI Architect at QualityKiosk Technologies · 8+ years in software engineering and architecture

I take AI agents from demo to production. That means the harness around the model (tools, skills, memory, permissions and human approval gates), the evals that gate every release, and the QA that keeps the whole thing honest. Before this I was Project Manager for AI capabilities on Katalon, and before that I built web and mobile products end to end at start-ups. On the side I run Releaslyy, an AI release-notes SaaS.

Skills

AI and agents

Agent harnessesSkills and subagentsMCPEvals and release gatesRAG on vector DBsToken optimisationModel routing

Quality engineering

Test automation strategyPlaywrightKatalonAI-assisted exploratory testingLLM test generation

Engineering

TypeScriptReactReact NativeNode.jsSupabase

How these posts are made

  1. RESEARCHSourced from public posts, docs and repos. Every claim links to its source.
  2. DRAFTWritten with AI from that research, in my voice, for each platform.
  3. REVIEWI edit and approve every word before it ships. Nothing posts on its own.

Elsewhere

Latest writing

All posts
  1. An agent harness checklist: tools, memory, permissions, approval gates

    5 min#agentsOct 28, 2026
    Oct 28, 2026
  2. Your CLAUDE.md is probably costing you more than your prompts

    3 min#claudecode#llmcostFirst on releaslyy.comJul 19, 2026
    Jul 19, 2026
  3. Why I built Releaslyy — the problem every sprint team ignores

    3 min#releasenotesFirst on releaslyy.comApr 2, 2026
    Apr 2, 2026