Bridging Quality and AI
The missing discipline in AI engineering: Quality.
I am Sourav Singh. I spent 10 years making software reliable. Now I am doing the same for AI systems, and helping QA engineers lead the shift rather than survive it.
Open source agent
Local-first AI agent by Sourav Singh / Sourav AI Labs.
Mini Krill runs with Ollama for private local use and can switch to subscription-backed Codex or Claude Code through their official CLIs. It is being hardened for reminders, email summaries, and unified chat memory across terminal, TUI, Telegram, and Discord.
curl -fsSL https://raw.githubusercontent.com/srvsngh99/mini-krill/main/scripts/install.sh | bashMay 01, 2026
Mar 04, 2026
Feb 25, 2026
Feb 22, 2026
Feb 20, 2026
Feb 18, 2026
Feb 11, 2026
Feb 04, 2026
Jan 28, 2026
Jan 21, 2026
12 posts
May 01, 2026
Mini Krill is now open source — a local-first AI agent built in Go that thinks, plans, and acts on your machine. Here's why I built it and what I learned.
Mar 04, 2026
Tokens, temperature, training vs inference. The conceptual foundations that change how you write test assertions for AI systems.
Feb 25, 2026
pytest advanced: fixtures, scopes, parametrization, conftest.py. The tools that make AI evaluation test suites actually work.
Feb 22, 2026
A lightweight framework for validating AI-driven product features.
Feb 20, 2026
What this site is about and what you can expect here.
Feb 18, 2026
Learning pytest from scratch and applying it to six weeks of accumulated code. Good test naming is documentation.
Feb 11, 2026
Building a robust API caller with retry logic and exponential backoff. LLM APIs fail more than you expect.
Feb 04, 2026
Reading and writing files, JSON serialization, and the JSONL format that underpins most AI evaluation datasets.
Jan 28, 2026
The week Python scripting became Python engineering. Building a reusable test utilities package from scratch.
Jan 21, 2026
Building a validation library for LLM outputs. Deterministic checks for non-deterministic systems.
Jan 14, 2026
Learning Python data structures by building a test case manager. The list-of-dicts pattern shows up everywhere in AI testing.
Jan 07, 2026
Starting from zero Python to building a reusable prompt template engine in one week. Including one hour lost to PYTHONPATH.
Mini Krill is now open source — a local-first AI agent built in Go that thinks, plans, and acts on your machine. Here's why I built it and what I learned.
May 01, 2026
Tokens, temperature, training vs inference. The conceptual foundations that change how you write test assertions for AI systems.
Mar 04, 2026
pytest advanced: fixtures, scopes, parametrization, conftest.py. The tools that make AI evaluation test suites actually work.
Feb 25, 2026
A lightweight framework for validating AI-driven product features.
Feb 22, 2026
What this site is about and what you can expect here.
Feb 20, 2026
Learning pytest from scratch and applying it to six weeks of accumulated code. Good test naming is documentation.
Feb 18, 2026
Building a robust API caller with retry logic and exponential backoff. LLM APIs fail more than you expect.
Feb 11, 2026
Reading and writing files, JSON serialization, and the JSONL format that underpins most AI evaluation datasets.
Feb 04, 2026
The week Python scripting became Python engineering. Building a reusable test utilities package from scratch.
Jan 28, 2026
Building a validation library for LLM outputs. Deterministic checks for non-deterministic systems.
Jan 21, 2026
Learning Python data structures by building a test case manager. The list-of-dicts pattern shows up everywhere in AI testing.
Jan 14, 2026
Starting from zero Python to building a reusable prompt template engine in one week. Including one hour lost to PYTHONPATH.
Jan 07, 2026
Walkthroughs, applied testing ideas, and updates from my public GenAI learning journey.
Feb 25, 2026
How to design build gates that protect release confidence without slowing teams down.
Feb 24, 2026
A practical foundation checklist for reliable AI-enabled product releases.