---
title: Tell me about the hardest problem you've ever solved.
description: Interview · Problem-Solving · Question 12
section: mind
tags: [interview, problem-solving]
genre: reference
stability: stable
lastUpdated: 2026-04-17
url: https://fardiniqbal.com/docs/mind/interview/problem-solving/tell-me-about-the-hardest-problem-youve-ever-solved
---


The mastery detection engine in Tiny Thoughts.

In ABA therapy, a child works on specific skills through discrete trials. A therapist records whether the child got it right, and at some point the child "masters" the skill and moves on. The question is: when? That determination is clinical judgment. BCBAs, the supervising therapists, look at patterns across sessions and decide. Different programs have different criteria. Different prompt levels within programs have different thresholds. There's no universal formula.

My job was to encode that judgment into software.

I spent weeks with BCBAs understanding how they actually make the call. Not what the textbook says. What they do. I learned there are three program types: Regular, Mand, and Eye-Contact. Each has its own prompt sequences. Mastery criteria vary per program type and per prompt level. There's a three-tier config cascade: organization defaults, template overrides, and individual enrollment overrides.

I built a pure TypeScript domain kernel. Configurable criteria at every level. Property-tested with fast-check to catch edge cases I couldn't think of. No machine learning. This problem doesn't need ML. It needs deep domain knowledge translated into deterministic rules with well-defined boundaries.

The engine has detected 129 mastery events. Each one means a real child demonstrated consistent ability in something they couldn't do before. The system recommends phase changes, suggests prompt fading. Therapists trust it because it works the way they think, just faster.

The hardest part wasn't the code. It was the domain learning. I had to understand ABA therapy well enough to write software that clinical professionals trust with their patients. That required humility, a lot of listening, and the willingness to say "I don't understand, explain it again" more times than was comfortable.