Selected case studies

Platform engineering for autonomous vehicles and production systems.

I turn operational bottlenecks into reliable software platforms—from product definition and architecture to hands-on implementation and production rollout. My recent work connects vehicle, backend, factory, and fleet workflows at meaningful scale.

13+years building production systems
4+years in autonomous driving
5,000vehicles supported by delivery infrastructure
30+vehicles calibrated per factory day

View résumé Email me

Core strengths

  • Vehicle delivery systemsSOTA/FOTA, maps, configurations, version state, rollback, and unreliable-network recovery
  • Operational platformsFactory calibration, incident management, route delivery, and cross-team workflows
  • Backend & systems engineeringPython, Linux, distributed services, messaging, observability, automation, and reliability
  • End-to-end ownershipProduct definition, architecture, implementation, deployment, and production iteration

Professional work

Featured projects

Internal product details are intentionally limited; the case studies focus on scope, decisions, and outcomes.

Rino.ai 2022–Present

Vehicle Software Delivery Platform

Fleet-scale SOTA/FOTA orchestration for autonomous vehicles

Vehicle Software Delivery Platform interface

My scopeArchitecture owner and lead platform engineer in a three-person vehicle-platform team

Challenge

R&D, factory, and fleet teams relied on separate delivery tools while vehicle count, software complexity, and map size were growing. Updates needed to remain observable and recoverable across unreliable networks, constrained vehicle storage, and distributed TBOX/ADCU nodes.

What I delivered

  • Defined the product direction and architecture for reusable vehicle-delivery infrastructure.
  • Architected and implemented a distributed FOTA MVP spanning orchestration, vehicle clients, backend services, web management, failure recovery, and power-loss safety.
  • Designed delivery strategies for software, configuration, and map datasets reaching hundreds of gigabytes.
  • Added resumable transfers, service-health monitoring, deployment automation, rollback safety, and version/state consistency.
30 → 10 min Average software deployment time
200 → 5,000 Supported vehicle deployment capacity
  • Python
  • Flask
  • React
  • Next.js
  • Celery
  • RabbitMQ
  • MQTT
  • MySQL
  • Linux
Rino.ai 2022–Present

Vehicle Calibration Platform

One calibration workflow for production lines and field operations

Vehicle Calibration Platform interface

My scopeEnd-to-end owner and lead developer

Challenge

Sensor calibration was slow, fragmented, and communication-heavy. The process limited factory throughput and forced field teams to spend up to a full day completing work on one vehicle.

What I delivered

  • Built an automated workflow covering calibration execution, validation, and operational handoff.
  • Integrated Python and Shell orchestration with C++ and OpenCV-based calibration capabilities.
  • Designed the platform for both repeatable factory production and fast field maintenance.
2–3 → 30+ Vehicles calibrated per factory day
0.5–1 day → under 3 min Field calibration time per vehicle
  • Python
  • Shell
  • C++
  • OpenCV
  • Linux
Rino.ai 2022–Present

Fleet Operations Platform

Incident management and route delivery connecting R&D with operations

Fleet Operations Platform interface

My scopeEnd-to-end product and full-stack engineer

Challenge

Autonomous-driving incidents and route delivery crossed team boundaries, but fragmented workflows made ownership, progress, and operational feedback difficult to track.

What I delivered

  • Built a shared workflow for incident intake, investigation, resolution, and delivery visibility.
  • Connected R&D and operations around consistent ownership, status, and route-delivery information.
  • Delivered the backend, frontend, data model, and production workflow as one integrated platform.
13 → 7 Incidents per 10,000 km within six months
R&D ↔ Operations Shared workflow and delivery visibility
  • Python
  • Flask
  • React
  • Ant Design
  • Tailwind CSS
  • MySQL
Dedao 2017–2022

Commerce & Finance Backend Platforms

Reliable order, inventory, settlement, and financial-recognition services

My scopeBackend engineer responsible for service design and delivery

Challenge

Commerce and finance teams needed dependable integrations across internal systems and major marketplaces, while manual office workflows created operational cost and reconciliation risk.

What I delivered

  • Designed backend systems for inventory, orders, distribution, and marketplace settlement across JD.com and Alibaba.
  • Built financial recognition, settlement, and analytics services that replaced manual office workflows.
  • Improved scalability and integration using asynchronous processing, event streams, GraphQL, TDD, and contract testing.
Automated Order ingestion and marketplace settlement
Replaced Manual finance recognition workflows
  • Python
  • Django
  • Flask
  • Go
  • Gin
  • Celery
  • RabbitMQ
  • Kafka
  • GraphQL
Kaoshixing 2014–2017

Kaoshixing B2B SaaS Platform

Zero-to-scale online examination platform for enterprise customers

Kaoshixing B2B SaaS Platform interface

My scopeArchitecture, full-stack development, DevOps, and engineering coordination

Challenge

Build a production-ready online examination product from the ground up while supporting diverse enterprise use cases, reliable delivery, and rapid customer growth.

What I delivered

  • Led system architecture and hands-on frontend and backend development from the initial product build.
  • Owned deployment, operations, maintenance, and engineering coordination as the platform scaled.
  • Supported enterprise training, recruitment, certification, practice, and knowledge-assessment workflows.
0 → 10,000+ Enterprise customers served
End to end Product, engineering, and production ownership
  • Python
  • Flask
  • JavaScript
  • MySQL
  • MongoDB
  • Linux