Xudong Chen

About

AI/full-stack engineer with experience delivering production-critical healthcare AI at Roche (digital pathology imaging platform using Microsoft Gigapath for cancer diagnosis) and building scalable LLM-backed products. Strong in designing async AI backends (Django/Celery/Postgres/Redis on GCP), reliability engineering (Datadog, incident response), and agent-style document analysis workflows with evaluation loops.

Experience

Education

Key Strengths

• Built and shipped production-critical medical imaging AI feature end-to-end (backend + frontend) for cancer diagnosis
• Full-stack delivery: designed Django/Postgres backend APIs and Angular/TypeScript frontend for initial release
• Designs scalable AI backends with async isolation (Celery), caching/rate limiting (Redis), and flexible structured storage (Postgres JSONB)
• Strong reliability/operations: debugged 5xx latency incident using Datadog traces/metrics; mitigated via rollback and rate limiting; added canaries, alerts, and circuit breakers
• Performance engineering via Locust load testing and failure injection; tuned worker concurrency, connection pooling, autoscaling, and retry/circuit-breaker behavior
• Pragmatic LLM product execution: scoped MVP with strict latency/cost constraints, deterministic JSON outputs, and audit logging; iterated using real usage metrics
• Agent-style AI workflow design with tool orchestration, persisted state, and evaluation loops (heuristics + human review)