Furqan Ali

Code in Place AI Evaluation Project

I was a named contributor on a peer-reviewed paper accepted at SIGCSE TS 2026, the leading ACM conference on computer science education. Our work, “Aligning Small Language Models for Programming Feedback,” demonstrated how rubric-based human preference signals could cut the performance gap to GPT-4.1 from 80% to 10%.

Link: https://sigcse2026.sigcse.org/

Capital One — Launchpad Leadership Program

I engineered a Python/Pandas analytics pipeline that surfaced behavioral friction points from high-volume user event logs. My team placed 3rd out of 16, and my work directly contributed to a 20% improvement in onboarding conversion.

I designed a structured data processing architecture that transformed raw behavioral data into actionable product insights, and presented findings to program leadership — translating complex data into business decisions.

Everyone Can Code (SWE)

I shipped a production NLP-powered application using the ChatGPT API that personalized AI-driven responses for user-generated content. The result: 25% increase in engagement and 30% increase in user retention, with 90% positive feedback from users.