Delivered faster, more reliable, more usable embedded software for Lumafield's X-ray CT scanners, enabling high-volume inspection usecases in factories.

• Reduced per-image processing time 30-50x to enable minute-long scans with quality equivalent to hour-long scans, replacing Scipy/Numpy/OpenCV algorithms with simpler, custom SIMD C code

• Improved scanner UI responsiveness 90% while making UI code easier to change and increasing type safety, by writing a Typescript code generator and implementing O(log(n)) instead of O(n) state updates

• Led 75% variable cost reduction project to move the 3d reconstruction pipeline from AWS servers to on-scanner CPU+GPU compute for high-volume customers

• Root-caused intermittent 300ms system stalls using full-system profiling and reverse engineering tools including NSight Systems, LD_PRELOAD, Cutter, and jemalloc

• Worked closely with QA, scanner fleet services, and frontend/backend software teams to implement, troubleshoot, and fix features for customers and internal teams

Developed volumetric renderers, hardware drivers, and server backends for accessible industrial inspection using X-ray CT.

• Implemented the core WebGL2 volume renderer customers use to look inside volumetric scans

• Developed the hardware driver for the in-house motion system board used in all scanners

• Created systems for version migrations and fleet-wide hardware health monitoring

• Designed and implemented scan data ingestion and security for the company's cloud server backend

• Produced detailed in-house documentation and presentations on software system performance, systemwide profiling, reverse engineering, and writing code that is easier to understand and change

• Improved the intern hiring process with individualized feedback

Create the "brains" for interactive garments and installations backed by Adidas, Steelcase, and Chicago's Museum of Science and Industry; and featured by outlets including WIRED, CNN, The Guardian, BBC News, Engadget, and CNET.

• In under 130 hours, engineered complete simulation suite, control system, and wifi monitoring portal for display case that reacts to human emotions. Project presented to executives at Adidas headquarters.

• Built visual calibration tool for tuning servo motors in 30 seconds (despite no prior servo experience).

• Achieved 20x performance boost for generative animations using game development tricks.

Designed, implemented, and documented the the render stack, IMGUI, and data tracing for an in-browser node-based programming environment and worked on the LLVM backend. Company acquired by Figma and then Adobe.

Led initiatives to unify user experience of revolutionary “live” programming environment for backend programmers, with goal of making coding 100x easier.

• Developed, documented, and tested standard toolkit with which customers build software used by 10s of thousands of end users.

• Introduced new programming paradigms and fixed biggest pain points during regular customer interactions.

• Became go-to person to fix most complex problems.