Vector Nguyen

About

Hi, I'm Vector Nguyen — AI Engineer at MoMo, based in Ho Chi Minh City. I specialize in multi-agent systems, LLMs, agentic voice, and production AI infrastructure. Background in Mechatronics Engineering (HCMUTE), transitioned into AI in 2022.


Experience

MoMo · AI Engineer Dec 2024 – present
Designed and shipped Tro Ly An Choi from zero — a multi-agent chat assistant (1 orchestrator + 6 domain agents) integrated across 10+ MoMo mini-app surfaces, reaching over 100K monthly engaged users. Built 70+ agent tools covering cinema, airline, bus, train ticketing, and food ordering with deep payment integration into MoMo's checkout system. Won Champions of Hacking Week at Tech Day MoMo 2025.

FPT Software – AI Center · AI Engineer Apr 2024 – Dec 2024
Built multimodal RAG pipelines, multi-agent architectures, and multilingual AI solutions for enterprise clients across Southeast Asia.

FPT Software – Quy Nhon · AI Engineer Jun 2022 – Apr 2024
Built computer vision and NLP systems. Received Most Valuable Player 2022 award. Won Second Prize at Quy Nhon AI Hackathon 2022.


Skills & Stack

Languages Python, TypeScript, JavaScript, SQL
AI / LLMs Multi-agent systems, RAG, prompt engineering, MCP, LLM evaluation, voice AI (ASR/TTS/VAD)
Frameworks FastAPI, Next.js, React, LangChain, LangGraph, Pydantic, WebSocket, MQTT
Databases PostgreSQL, pgvector, Elasticsearch, Redis, Qdrant, BigQuery
Observability Langfuse, Grafana
Cloud & DevOps AWS (Lambda, S3, CloudFront, EKS), GCP, Docker, Kubernetes, Airflow, GitHub Actions, Nginx

What I'm Building

Outside of work I build Assistant Core (assistantcore.com) — a platform for deploying one AI assistant configuration everywhere: web, mobile widgets, IoT devices, wearables, and embedded surfaces, without duplicating logic per channel.

  • Multi-provider AI: supports 7+ providers (OpenAI, Anthropic, Google, xAI, DeepSeek, AWS Bedrock, ElevenLabs, Soniox) — switchable without reconfiguration
  • Voice pipeline: sub-1s latency ASR → LLM → TTS streaming with VAD auto mode, barge-in interruption, wake-word customization, WebSocket and MQTT+UDP transport
  • Knowledge & tools: agentic RAG over documents, websites, and databases; MCP server integration; custom API tool calling
  • White-label multi-tenant: RBAC, custom domain routing, full-featured admin dashboard for assistant config, user management, device pairing, and voice quota controls
  • Stack: PostgreSQL/pgvector, Redis, S3, Langfuse tracing, Grafana monitoring, TLS 1.3, JWT/OAuth 2.0

How I Work

I care about shipping real products — not prototypes. My instinct is to go from zero to production as fast as possible, then iterate on real user feedback. I tend to think in systems: how components connect, where failures propagate, what the user actually experiences end-to-end. I'm most energized when I can own a problem fully, from architecture to deployment.

Feel free to reach out on LinkedIn or email to collaborate.