A living snapshot of what I am building, teaching, writing, and helping organize, not the complete historical record.
This month I spoke as a panelist at the IEEEXtreme 20.0 Masterclass, ran Columbia’s in-person sitting of the 2026 North America Qualifier, and continued the weekly masterclass and Online Training in Competitive Programming loop for ICPC Foundation interns.
The common thread is straightforward: put serious work in front of people, make the feedback immediate, and leave them better able to own the next problem.
On October 10, I joined the IEEE Student Activities Committee’s five-hour global masterclass as a panelist. It was a chance to work directly with the kind of audience I care about most: people preparing to solve unfamiliar problems under real time pressure.
I organized Columbia’s October 3 sitting with 117 individual registrations and continued as an NAQ judge. Across North America, the contest drew 1,575 teams from 161 schools.
As Chief Architect at Probook, my current work is the architecture around real-time AI for home-service businesses: speech, model orchestration, tool use, workflow state, observability, and the recovery paths that make an agent useful outside a demo.
The interesting problems sit at the boundaries: between a caller and a model, between a model and a business system, and between a plausible response and a task completed correctly.
I am also applying an operations-research discipline to production decisions: define the operating objective, measure the real tradeoffs, test assumptions under adverse conditions, and choose the least complex approach that clears the bar. The same ownership extends through the full lifecycle, including deliberate retirement, cost visibility, knowledge preservation, and clear handoffs when a system should no longer remain active.
At the organizational level, I am also making technical direction easier to act on: clearer decisions, earlier risk visibility, stronger alignment across engineering leads, and more independent ownership throughout the team.
I am preparing upcoming offerings in competitive programming and technical-interview preparation while continuing to coach Columbia’s competitive-programming community.
The goal is not just to expose students to more algorithms. It is to help them develop a repeatable process for reading unfamiliar problems, testing ideas, debugging reasoning, and reviewing practice.
At the ICPC Foundation, I am turning the internship initiative from a successful effort into a durable program: formal governance, explicit outcomes, repeatable operating workflows, shared ownership, and succession. Intern feedback currently averages 9.6/10 across 86 responses.
I am also helping build ICPC University into a useful learning commons, with structured routes across curriculum, practice, lectures, camps, events, and contest operations for students, coaches, educators, and organizers.
Josh Alman and I are developing Competitive Programming: Beyond the Basics. The working manuscript now spans twenty chapters, with a classroom edition supporting test lectures and revision.
I am also publishing shorter working notes about AI systems, practice, mentorship, and technical education.
Real-time AI and agent architecture; invited talks and guest teaching; competitive-programming programs; and thoughtful questions from students choosing their next step.