Engineering Intelligence, Not Just Prompts
I am an AI Engineer at HPT Vietnam Corporation (AI Center of Excellence). My engineering focus is bridging the gap between fragile proof-of-concepts and deterministic, enterprise-grade AI systems.
I architect production systems that operate with mathematical grounding, sub-second streaming latencies, and zero-hallucination guarantees. My core toolkit centers on Agentic RAG, autonomous tool-routing via Model Context Protocol (MCP), hybrid semantic search (Qdrant), and low-latency voice/text pipelines. Rather than treating LLMs as magic black boxes, I design the robust distributed plumbing around them — deterministic routing, strategic embedding chunking, self-healing retrieval loops, and rigorous evaluation harnesses.
Key Milestones & Production Track Record
-
Architected "Lyly" (Executive AI Assistant): Built a multimodal assistant with bi-directional voice & text streaming via WebSockets, hybrid vector search (Qdrant), and dynamic contextual memory (
mem0). Awarded 2nd Prize in HPT's Enterprise AI Competition.
-
Orchestrated HPT D-DAY 2025 RAG Chatbot: Deployed a high-throughput, zero-hallucination RAG chatbot handling 850+ live executive queries in a single day at Pullman Hanoi with 100% grounded accuracy against official knowledge bases.
-
Co-engineered HIH ("Sóc Dò La"): Automated intelligence pipeline vectorizing and synthesizing industry trends into actionable executive briefings via Microsoft Viva Engage (8,500+ impressions, 3rd Prize Winner).
-
Enterprise Copilot & Process Automation: Formulated advanced grounding and prompt optimization frameworks for Microsoft Copilot deployments, paired with 10+ production workflows (n8n, Power Platform) cutting manual cycles by ~60%.
I thrive in fast-moving engineering environments that demand high velocity and architectural rigor. I'm currently open to AI Engineer and Agentic Systems Architect roles where I can own the complete lifecycle from research to scaled production.
.webp)
.webp)
