MUHAMMAD IBRAHIM

Joined a research team studying correlations between tumor size and eye occlusions to get closer to the intersection of software and healthcare. Taught myself how medical imaging models work from the ground up, from preprocessing and segmentation to evaluation in clinical contexts where a false negative carries real weight. Completed Health Canada GCP and TCPS 2 certifications to be qualified to participate. What started as curiosity became a north star: when nobody tells me what to work on, I work on healthcare AI.

Worked alongside a senior engineer to build an NLP-based script analysis tool that helps filmmakers estimate tax credit eligibility — a process that used to require consultants or hours of manual research. When the MVP wasn't telling a compelling enough story for investors, I took ownership of rearchitecting core features, tightening the user flow, and getting a demo-ready version finished before the Y Combinator deadline.

Graded Python assignments and held weekly office hours for 200 students. The real value was learning how to explain technical concepts clearly and meet people where they are, a skill that shows up in every cross-functional conversation I have now. Mentoring students on their term-long projects taught me as much about breaking down problems as it taught them.

Built a Python network analysis tool to evaluate LEO satellite system performance. My analysis scripts ran painfully slow until a fellow intern asked why I wasn't using async functions — that conversation taught me something I still carry: when you're stuck, describe your problem to someone else. The fix often comes not from their answer but from the new frame they give you.