Building intelligent products from model to market

|

I design, train and deploy AI experiences that feel fast, useful and production-ready. From LLM agents to MLOps platforms, every interface is backed by measurable intelligence.

10+
AI products shipped
1M
predictions per day
1 yrs
ML engineering
LLM
RAG
MLOps
CV
DE
BE

About

Bridging research and product with production discipline.

I am an AI engineer who turns ambiguous ideas into shipped systems. I have spent the last 1 years building LLM applications, computer-vision pipelines and MLOps platforms for startups and scale-ups. Besides, I have also worked as a Backend Engineer and Data Engineer.

Clarity first

Complex models are useless unless the product around them is understandable.

Reliability by design

Guardrails, evals and observability are built in from day one, not bolted on.

Fast iteration

Rapid prototypes with measurable feedback loops keep teams moving and learning.

Outcome obsessed

Every feature is traced to a business metric: latency, accuracy, revenue or retention.

From notebook to production traffic.

My work sits at the intersection of machine-learning research, software engineering and product design. I enjoy the full arc: framing the problem, collecting signal, training or fine-tuning models, building the serving layer, and refining the UX until users trust it.

Whether it is a retrieval-augmented agent, a real-time vision pipeline, or a forecasting service, I ship systems that teams can depend on.

AI, DE, BE

Skills

A toolkit built for end-to-end AI product delivery.

Deep expertise where models meet infrastructure, interfaces and measurable outcomes.

AI & Machine Learning

LLM SystemsRAGAgentsFine-tuningEval PipelinesComputer VisionTime SeriesReinforcement Learning

Engineering & MLOps

PythonPyTorchFastAPILangChainVector DBsKafkaAWS

Product & Data

FastAPIPostgresFeature StoresWeights & BiasesMySQL

Selected Work

Production AI with a polished product surface.

Case studies combine robust model engineering, clear UX and measurable outcomes for real teams.

Agentic RAG
01

Agentic AI Trading System CLI

A multi-agent knowledge assistant that resolves enterprise tickets with citations and live tool use. It is a CLI tool that allows users to interact with the system through a terminal.

63% faster resolution
RAG
02

Multi-Agent Conversational Sales Copilot with Hybrid GraphRAG

A multi-agent conversational sales copilot that allows users to interact with the system through a voice interface. It is a copilot that transcribes the user's voice to text, then uses the text to generate a response.

91.4% defect recall
Voice-to-Voice
03

Real-time Voice-to-Voice AI Pipeline

A real-time voice-to-voice AI pipeline that allows users to interact with the system through a voice interface. It is a pipeline that transcribes the user's voice to text, then uses the text to generate a response.

100% accuracy

Experience

A track record of shipping AI at scale.

Roles that required both deep technical ownership and cross-functional product leadership.

2022 — 2026

Student

Saigon University

I'm an IT engineer specializing in Artificial Intelligence from Saigon University (SGU). With the discipline of a Bodybuilder 🏋️ and the sharp mindset of a Crypto Trader 📈, I approach code and data architecture with a focus on performance and optimization.

  • AI Engineer
  • Data Engineer
  • Backend Engineer

Blog

Notes on building with AI.

Practical write-ups on agents, evaluation, infrastructure and product design.

LLM Systems

RAG beyond retrieval: designing agents that actually know when to act

Why retrieval is only half the battle, and how planning, tool use and self-evaluation make agents trustworthy.

8 min read
Evaluation

Evaluating LLM agents: a practical framework for product teams

From unit tests to human-in-the-loop ratings: a layered approach to measuring agent quality in production.

6 min read
MLOps

MLOps for small teams: staying lean without cutting corners

The minimum viable tooling stack for versioning, monitoring and deploying models with confidence.

7 min read

Contact

Have a model, dataset or product idea ready to become real?

Tell me what you are building and I will get back to you within two business days.

Prefer email? tientho2012004@gmail.com