Hi, I'm Dev. I work on
distributed systems
and platform engineering.
Recent graduate of the University of Illinois with a combined Bachelor's and Master's in Computer Science. I work on distributed systems and platform engineering.
I recently graduated from the combined Bachelor's/Master's in Computer Science program at UIUC, with a minor in Business. Most recently I interned on Capital One's luxury travel technology team in New York, building a serverless data pipeline for generative travel itineraries. Before that I worked across fraud platforms, MLOps at Caterpillar, and security tooling.
- Working with commerical clients.
- Integrated a travel itinerary generation solution using hotel and flight data sources.
- Built a serverless pipeline sharing travel context with a GenAI provider while protecting customer privacy.
- Unified automated and manual fraud decisioning into one platform with shared-attribute linking.
- Cut average handling time 16%, saved $5M in operational costs, improved investigation throughput.
- Eightfold throughput gain via event-driven architecture, async I/O, multi-threading, and Redis caching.
- Cut anomaly detection pipeline runtime by 86%; pruned obsolete prediction models from S3.
- Shipped open-source Canvas/QTI → PrairieLearn converter; designed and merged question preferences.
- Rebuilt distributed autograder on Jenkins, ran monitoring stack, served on-call for production incidents.
- Built full-stack tooling that cut non-compliant access to regulated systems by 15%.
- Shipped an end-to-end encrypted file transfer portal and led NIST/OCTAVE security audits.
A few projects I've worked on.
Parser that converts Canvas/QTI exports into PrairieLearn's native format, automating assessment migration that used to be fully manual. Shipped upstream into PrairieLearn, the open-source platform used across many universities.
A merged feature adding parameter passing to PrairieLearn questions, so one question can be safely reused across assessments with different configs — designed, reviewed, and shipped into a large production open-source codebase.
Distributed, fault-tolerant autograder using a Jenkins-based runner for UIUC's Systems Programming course — faster and more reliable, with richer feedback for the hundreds of students who depend on it each semester.
Built the platform behind ACM @ UIUC's operations — auth, membership, and events. Engineered for high availability with a multi-region, serverless AWS architecture and automated failover that keeps it reliable for thousands of members.