52 Weeks of GenAI Testing on Sourav AI Labs

Week 9: How LLMs Work, and Why That Changes How We Test Them

Wed, 04 Mar 2026 00:00:00 +0000

Week 9 of 52 Phase 1: LLM Fundamentals Status: In Progress

Phase 1 starts here. Eight weeks of Python and testing tooling, now applied to the thing I actually care about: understanding how language models work from the ground up, so I can test them properly.

Week 8: Fixtures, Parametrization, and Testing at Scale

Wed, 25 Feb 2026 00:00:00 +0000

Week 8 of 52 Phase 0: Foundation Status: Complete

Week 8 completed the Foundation phase. The topic: pytest advanced features. Fixtures, fixture scopes, parametrization, conftest.py, and markers.

This is the week where writing tests stops being tedious and starts being powerful.

Week 7: pytest Basics and Why Test Discovery Matters

Wed, 18 Feb 2026 00:00:00 +0000

Week 7 of 52 Phase 0: Foundation Status: Complete

Week 7 was the first explicitly testing-focused week of the journey. After six weeks of building Python tooling, it was time to write proper tests for it.

Week 6: Error Handling, Logging, and the Unreliable API Problem

Wed, 11 Feb 2026 00:00:00 +0000

Week 6 of 52 Phase 0: Foundation Status: Complete

Week 6 was about building software that fails gracefully. The topic: try/except/finally, custom exceptions, and the Python logging module.

The mini-project made the stakes concrete. LLM APIs are not like database queries. They time out. They rate limit. They return partial responses. They fail in ways your happy-path tests will never catch.

Week 5: File Handling, JSON, and Why JSONL Is the Format for AI Testing

Wed, 04 Feb 2026 00:00:00 +0000

Week 5 of 52 Phase 0: Foundation Status: Complete

Week 5 was practical infrastructure: reading and writing files, JSON serialization, pathlib for modern path handling, and JSONL for large datasets.

Sounds dry. It is not. If you test AI systems, you live in JSONL.

Week 4: Modules, Packages, and Writing Code That Lasts

Wed, 28 Jan 2026 00:00:00 +0000

Week 4 of 52 Phase 0: Foundation Status: Complete

Week 4 is when things started feeling like real engineering rather than learning exercises.

The topic was modules, packages, and virtual environments. The skill: structuring code so other code can use it.

Week 3: Control Flow, Functions, and Validating LLM Responses

Wed, 21 Jan 2026 00:00:00 +0000

Week 3 of 52 Phase 0: Foundation Status: Complete

Week 3 was where things started connecting to the actual problem I care about. Control flow and functions are foundational Python - but the mini-project applied them directly to LLM output validation.

Week 2: Data Structures and the List-of-Dicts Pattern

Wed, 14 Jan 2026 00:00:00 +0000

Week 2 of 52 Phase 0: Foundation Status: Complete

Week 2 was data structures: lists, dicts, sets, tuples. The goal was not to memorize syntax but to understand which container to reach for and when.

Week 1: Python Basics, Prompt Templates, and a Debugging Rabbit Hole

Wed, 07 Jan 2026 00:00:00 +0000

Week 1 of 52 Phase 0: Foundation Status: Complete

The first week was about proving something simple: that a QA engineer with 10 years of experience but minimal Python could get moving fast when the goal was concrete.