Lead Agentic AI Engineer
Leads a small AI team delivering LangGraph-based agents and RAG systems for enterprise clients, with production discipline around LLM-as-judge evaluation, golden datasets and Langfuse/OpenTelemetry tracing. Also brings deep computer-vision and document-intelligence experience, including OCR for complex business forms and model compression for edge and real-time use.
- Level
- Lead
- Experience
- 4–6 years
- Based in
- Vietnam
- Languages
- Vietnamese (native), English (B2 intermediate)
Selected work
- Enterprise AI (Japanese clients): led delivery of a real-time digital-human assistant with domain RAG on AWS EKS, keeping responses around 2–3 seconds while shrinking the animation model about 10x through distillation and quantization.
- E-commerce / fashion retail: built a conversational RAG agent with hybrid search and a multi-tier LLM-as-judge evaluation suite over a 400+ case golden dataset, fully traced with Langfuse and OpenTelemetry.
- Back-office document processing: trained vision models that extract key fields from complex Japanese business forms at about 98% accuracy and ~2 s latency on edge hardware.
- HR / workforce planning: combined a constraint solver with an LLM on AWS Bedrock to turn policy documents into constraints and generate ranked leave schedules.
Domains
Enterprise AI, E-commerce & retail, Document processing, HR tech
Education
Bachelor's, Computer Science / AI