PinnedRaj Kundalia·Aug 26When Building an AI Agent, the Journey Matters as Much as the DestinationA practical framework for evaluating trajectories, tool use, and process, not just the final answer.A response icon1A response icon1
PinnedRaj Kundalia·Apr 26How I Review PRs with AI — Without Losing My Own JudgmentOver the last few months, my code review queue has changed completely. With agentic coding, PRs are larger, generated faster, and harder to…
PinnedRaj Kundalia·Jun 13Where You Put the Instruction Matters More Than What It SaysAn experiment comparing system prompts, user prompts, and tool descriptions across Claude and QwenA response icon1A response icon1
PinnedRaj Kundalia·Mar 21Pydantic vs Instructor vs BAML: Which One Actually Solves LLM Output Parsing in Production?TL;DR
Raj Kundalia·Jun 28What Happens When Every Prompt Slot Says Something Different?A controlled experiment exploring how Claude and Qwen resolve conflicting instructions across system prompts, user messages, and tool…
Raj Kundalia·Jun 7Why 95 Reviews Beats 20 Reviews — Even When Both Score 95%Understanding Wilson Score, confidence intervals, and the mysterious 1.96.
Raj Kundalia·May 12Why Your Story Points Feel Arbitrary (And How to Fix It)I just feel it is “2”, no, I think it is “3”.
Raj Kundalia·Apr 11Following a Database Read to the Metal — A Simple WalkthroughI wanted to learn about internals of indexes in database and the first step towards that was — how Disk IO works — I took help from…
Raj Kundalia·Feb 22From println to Production Logging: Internals and Performance Across Languages and the OSIf you do not want to read the article, it is A-OK:
Raj Kundalia·Jan 31Distributed Tracing in Spring Boot: A Practical Guide to OpenTelemetry and JaegerTL;DR