Training Rails
A rail-defect segmentation model with 95% detection success; Jetson inference and GPS detections stream to a live map.
ABOUT
Founding Engineer at Zyp, rebuilding the backend for a live cross-border payments platform. Previously a Software Engineer Intern at Next League, where I built the PostgreSQL/TimescaleDB data layer and REST API for an internal platform-health analytics product.
Computer science student at Oregon State University. At Warren Lab, I deployed a three-stage computer-vision pipeline to Raspberry Pi that sustained 20 FPS across 166 field sessions and more than 1.5 million frames.
A rail-defect segmentation model with 95% detection success; Jetson inference and GPS detections stream to a live map.
Built in 24 hours: dynamic routing across 100+ LLMs, SSE streaming, and ElevenLabs text-to-speech.
A pip-installable image-labeling CLI with persistent sessions, undoable changes, and remote labeling over SSH.
A relational database with joins, CTEs, window functions, and twelve progressive SQL scripts.
Founding Engineer · Part-time
AUG 2026 – PRESENTLeading an 8-milestone rebuild of a live payments platform after a Zyp internship designing its normalized PostgreSQL model and auditing backend authorization.
Mapped the legacy Supabase backend into eight milestones; own API development, authentication, and deployment.
Enforced tenant isolation with PostgreSQL RLS, separate roles, and a Prisma wrapper that prevents context carryover across pooled connections.
Built sessions and password recovery with Better Auth; shared Zod/OpenAPI schemas validate request bodies at startup and database roles before traffic.
Software Engineer Intern · Part-time
JUN 2026 – AUG 2026Built the PostgreSQL/TimescaleDB data layer and REST API for an internal platform-health analytics product.
Built the PostgreSQL and TimescaleDB foundation for operational and historical platform-health data.
Connected the data model to the internal analytics product with Payload CMS and Next.js 16.
Separated editable configuration from append-only snapshots, with transactional roll-up hooks.
Undergraduate Researcher · ML / CV
FEB 2025 – PRESENTDeployed a three-stage computer-vision pipeline to Raspberry Pi across 166 field sessions and more than 1.5 million frames.
YOLOv8, ResNet-50, and Kalman tracking sustained 20 FPS at 99.5% mAP@50.
Replaced about 20 seconds of manual labeling with 74 ms inference at 99.7% precision and found 200+ errors in a 10,000-image dataset.
Built a ResNet-18 orientation classifier with 100% on a 419-image held-out set; second author on a July 2026 bioRxiv preprint.
B.S. Computer Science · Sep 2024 – Jun 2028 (expected)
Dean's List · GPA 3.90 / 4.0 · Corvallis, OregonHigh School Diploma · Mathematics Concentration
St. Petersburg, Russia · Graduated 2024 · #50 nationally for STEM placement (RAEX 2024)Advanced computer science program · 2021
Graduated with a voice-controlled chess project.