What I build
- Domain-grounded assistants that reduce hallucination and support clear actions
- Real-time multimodal systems that handle high-stakes support workflows
- Cloud-native application stacks with measurable SLO and delivery ownership
Open to AI Engineer opportunities in Canada and the US
Building AI products that deliver measurable outcomes
I engineer production LLM assistants with strong reliability, clear business impact, and practical deployment strategy.
AI Software Engineer with experience across GPT-powered assistants, RAG pipelines, voice-first emergency systems, and enterprise cloud modernization.
About
Applied AI engineer focused on production reliability
How I work
- Start from business constraints, then design model + system boundaries
- Iterate with user feedback loops and grounded evaluation criteria
- Ship maintainable software: testability, observability, and rollback safety
Experience
Professional timeline
ERP System Software Engineer & Functional Consultant
JTANG Consulting Inc | Vancouver, BC, Canada
- Customized Microsoft Dynamics 365 Business Central for 10+ suppliers, improving operational efficiency by 30% and reducing inventory errors by 50%.
- Delivered end-to-end migration from legacy ERP to Azure-based Business Central, including full MySQL migration and production cutover.
- Resolved 50+ Tier-2 support tickets across access control, procurement, and configuration while maintaining high user satisfaction.
AI Software Engineer
We Link Technology LLC | San Jose, CA, USA
- Led development of JuTong, an AI sales and service assistant for KONE, using GPT-4o + domain retrieval to automate diagnostics and recommendations.
- Drove about 30% team efficiency improvement and contributed to $5M cost savings.
- Built backend services with Python/Flask, containerized via Docker, and deployed to AWS EC2.
- Implemented RAG over 500MB structured documents to raise grounded answer quality.
- Built a Twilio-based emergency voice platform and improved grounded accuracy from 80% to 90%, reducing escalations by 50%.
Research Assistant
UC San Diego SOC Lab | San Diego, CA, USA
- Developed an AI-based traffic control simulator in Python with 20% simulation efficiency gains.
- Built real-time simulation visualization and validated 15% faster traffic flow stabilization.
- Delivered a Flask demo app for research communication and result presentation.
Projects
Selected AI systems
Enterprise LLM Assistant
JuTong AI Sales & Service Assistant
Production AI copilot for elevator sales/service workflows. Delivers contextual recommendations, diagnostics support, and faster customer operations.
GPT-4oRAGPythonFlaskAWS EC2
Real-time Voice AI
AI Elevator Emergency Call Center
End-to-end voice assistant for emergency response with continuous passenger interaction, engineer relay, STT/TTS pipeline, and fallback resilience.
TwilioAzure AISTT/TTSGPT-4o Fallback
Applied Research
Autonomous Traffic Control Simulator
AI traffic strategy prototype with live simulation and web visualization for mixed highway traffic behavior analysis.
PythonFlaskSimulationVisualization
Skills
Technical stack
AI / ML
LLM ApplicationsRAGFine-tuning
Stable DiffusionPrompt Engineering
Languages
PythonJavaC++JavaScript
SQLHTMLCSSMATLABAL
Frameworks & Platforms
FlaskDjangoReactAWS
AzureDockerMySQLMongoDB
RedisSpark
Enterprise Delivery
Dynamics 365 Business CentralERP Migration
Tier-2 SupportRequirements Analysis
Education
Academic background
- University of California, San Diego MSc in Machine Learning & Data Science | GPA 3.6 / 4.0 Sep 2021 - Mar 2023
- Cardiff University BSc in Computer Science (First Honor) | GPA 4.0 / 4.0 Sep 2019 - Jun 2021
Awards
Recognitions
- Silver Award 6th China College Students Internet Plus Innovative Competition 2020
- National Project Approval Lizi Visitor Platform based on SaaS mode 2019
- Excellent + Most Popular Award Microsoft Imagine Cup competition 2019
Contact
Looking for an AI engineer who ships end-to-end systems?
I am available for roles focused on applied AI, LLM platforms, and production software engineering.