Shayan’s notebook / people, then systems
People, then systems.
I like working close to the problem: talking to the people using a system, building it, and staying around when it meets the real world.
Software engineer & forward deployment
Hamming AI
I built evaluation systems for voice agents, from making a single call easier to debug to finding problems across thousands of conversations.
Alongside product and infrastructure work, I led forward-deployed delivery for NVIDIA and xAI. Both engagements expanded after the initial delivery.
A closer look at what I built +
Automated classification
Turned raw calls into decision-ready labels teams could use immediately.
Cross-call analysis
Replaced isolated anecdotes with patterns visible across conversations.
Recurring issues
Converted repeated production failures into prioritized product work.
Hamming Assistant
Made production evidence directly queryable and easier to act on.
Evaluation workflows
Moved evaluations from initial setup into reliable, repeatable production use.
Trace exploration
Made complex call histories faster to inspect, compare, and debug.
Dataset operations
Improved how evaluation data is curated, reused, and carried between workflows.
Quality dashboards
Turned dense evaluation output into signals teams could read at a glance.
Failure triage
Connected production evidence to faster investigation and clearer next actions.
Production integrations
Brought evaluation capabilities into the environments where teams already worked.
Undergraduate research assistant
University of Waterloo
I worked with Professor Elena Grigorescu on space-efficient Euler-tour algorithms and heapable subsequences. The questions were about doing more with less memory, and finding the right argument for why it works.
USACO competitive programming mentor
Breakout Mentors
I coached 25+ students in algorithms and data structures, working through USACO Bronze, Silver, and Gold in C++, Python, and Java.
The interesting part is figuring out what someone sees when they read a problem, and helping them turn an intuition into an argument they can trust.



Event Tech Lead · Grebel Tech Team
Conrad Grebel University College
I managed A/V setup, event logistics, and live technical troubleshooting with the Grebel Tech Team, keeping university events running smoothly.
At the residence, I also built GrebelCup: a weekly trivia and challenge platform used by 100+ students.
Contest organizer & problem author
Codeforces
I co-authored Codeforces Round 1024, which drew more than 24,000 participants. Setting problems means thinking both like a solver and like someone trying to break every assumption in a solution.
On the other side of the contest, I’m a Codeforces Master, with a peak rating of 2221 and 1,797+ problems solved.
Software engineering intern
Web24 / SEO24
I worked on frontend development and page-load performance for a platform serving more than 100,000 monthly users. It was an early lesson in building things for people beyond my own screen.
“Don’t worry if it doesn’t work right. If everything did, you’d be out of a job.”Mosher’s Law of Software Engineering