Jaxon Wong
I’m Jaxon Wong, an AI/ML and full-stack engineer based in Hong Kong.
About
I’m Jaxon Wong, an AI/ML and full-stack engineer based in Hong Kong. Over the past 9 years, I’ve moved from mobile development into production-grade web systems and applied machine learning, with a focus on turning client workflows into reliable products—end to end, from model evaluation and API design to user-facing experiences.
I enjoy building permission-aware retrieval and document intelligence systems, offline-capable mobile apps, and structured-LLM pipelines that include validation, monitoring, and rollback-ready deployments. I’m happiest when I can connect rigorous experimentation (metrics, error analysis, experiment tracking) with clean engineering practices (testing, observability, and deployment automation) so teams can ship with confidence.
- 9 years' experience
- Joined October 2026
Industry experience
- Software & Internet
- Computers & Electronics
- Professional Services
Work experience· 4 roles
Led AI-enabled web and mobile engagements from discovery and technical design through implementation, deployment, and support. Built a permission-aware document assistant using FastAPI, pgvector, and Next.js, improving retrieval recall@5 from 71% to 88% on a 200-question evaluation set. Delivered an offline field-service React Native app with retry-safe synchronization and automated build, test, and deployment workflows.
Developed a support-ticket classifier using scikit-learn baselines and PyTorch, improving macro-F1 from 0.74 to 0.86 on a held-out test set. Analyzed time-based splits and per-class errors to assess generalization, and routed low-confidence predictions to manual review. Implemented experiment tracking and model versioning with MLflow; deployed Dockerized FastAPI inference services with validation, monitoring, and rollback procedures.
Built a React, Node.js, and PostgreSQL operations portal covering order workflows, role-based access, and audit history, plus background export jobs. Reduced p95 API latency from 820 ms to 310 ms in load tests using indexing, pagination, and Redis caching. Added integration tests for critical workflows.
Education
Hire Jaxon Wong today
Post a job and invite Jaxon, or contact them directly. No hidden commission.