About

I care about systems that fail loudly, APIs that are hard to misuse, and code the next person can read without reverse- engineering my intent. Complexity is a cost. I try not to spend it unless I'm getting something real back.

I also care a lot about recovery. A lot of engineering mistakes are survivable if the system makes them obvious and cheap to fix. Silent failures are the worst kind.

I like ambiguous problems when the constraint is real. At Opslyft that meant customer-facing architecture work, cost investigations, and shipping changes into environments I did not fully control.

I try to reduce ambiguity for everyone else: make the tradeoff explicit and the failure mode visible.

Before grad school, I spent close to three years at Opslyft working on cloud cost infrastructure, multi-cloud visibility, and reliability-sensitive backend systems. That's where I learned the difference between code that works and systems that hold up in production. Alongside the work, I co-presented data transfer cost optimization at Rootconf 2022.

Then I went to Purdue Fort Wayne for my Master's, where I worked on adversarial ML and Android malware detection under Dr. Zesheng Chen. That work became a peer-reviewed paper in MDPI Electronics.

Dr. Chen had one question he kept asking: "Why are you assuming that's the right approach?" That question changed how I think about almost everything.

Alongside the research, I TA'd undergraduate machine learning for two semesters. I automated parts of the grading pipeline, but the better part of the experience was explaining where the math, implementation, and model behavior actually meet.

Gym and books. The gym taught me something about consistency that I haven't been able to learn any other way.

The books I come back to are usually the ones that change how I think, not just what I know.