Skip to content

cv.md

Latest commit

 

History

History
72 lines (45 loc) · 4.76 KB

cv.md

File metadata and controls

72 lines (45 loc) · 4.76 KB
Raw

Quan Hua

Senior AI Engineer

Ho Chi Minh City, Vietnam | quanhua92@gmail.com
linkedin.com/in/huahongquan | github.com/quanhua92

Professional Summary

Senior AI Engineer (10+ years) specializing in multimodal AI systems, agentic LLM workflows, and production ML pipelines. Patent holder in multimodal sensor fusion (radar+optical) for 3D body measurement at BodiData (3 scanner generations), with prior deep learning work on edge robotics (Maya Personal Robot). Founded two open-source AI products: OpenVideoKit (AI-first video editor with a LangGraph copilot, MCP tool server, and neural TTS pipeline) and AIPriceAction (Rust data pipeline and Agent Skills). End-to-end ownership from model design to cloud/edge deployment with strong MLOps practices.

Experience

Oct 2025 – Present: Independent Founder — Open-Source AI Products

Building and shipping two open-source AI products end-to-end.

OpenVideoKit - https://www.openvideokit.com/

  • Built an AI-first video editor with a multimodal rendering pipeline (FastAPI → animated HTML → neural TTS → MP4) and low-latency SSE streaming.
  • Designed a LangGraph copilot that proposes human-gated edits, exposed via a 28-tool FastMCP server for tool-augmented LLM agents.

AIPriceAction - https://www.aipriceaction.com/

  • Engineered a Rust backend (Axum, PostgreSQL, Redis) with 22 workers syncing 574 global tickers, plus a multi-agent research pipeline with automated quality gates.
  • Shipped a React 19 + TypeScript SPA and a Python SDK + CLI (aipriceaction, aipa-cli) with 3 Agent Skills for Claude Code, Cursor, and OpenCode.

2017 – Oct 2025: BodiData, Inc - https://www.bodidata.com/ Senior Machine Learning Engineer

3D body measurement and size-matching solutions to reduce clothing returns. Led the multimodal CV engine across 3 generations of the Kora® 3D Body Scanner, fusing depth, RGB, and radar data into accurate body models.

  • Kora V1 (Surface + RealSense): Designed the multimodal CV pipeline from scratch — fused RealSense depth with RGB and radar via ORB-SLAM 3 for real-time camera positioning; implemented the C++ application in Qt and collaborated on the radar fusion algorithm.
  • Kora V2 (RealSense D415): Ported to far-range camera. Decreased 3D measurement error by ~1 inch through enhanced model fitting against fused point cloud data, improving product accuracy.
  • Kora V3 (iPad + ESP32 → AWS Lambda): Rewrote ESP32 firmware to multi-threaded ESP-IDF with BLE to iPad and Vayyar radar integration for depth fusion.
  • Kora V3 Cloud: Refactored Qt C++ processing pipeline into a CLI binary and built a Python Lambda to orchestrate S3 downloads, processing, and uploads.
  • Re-engineered Node.js SSE streaming server in Rust (Axum, Tokio), eliminating OOM crashes and boosting throughput 10x on smaller ECS instances.

2014 – 2017: Autonomous, Inc. - https://autonomous.ai/ Artificial Intelligence Engineer

  • Directed 6 AI engineers to build deep learning brain for Maya Personal Robot (CES 2015 TechCrunch Hardware Battlefield finalist), contributing to $161K+ raised on Kickstarter.
  • Integrated and deployed deep learning models for computer vision, facial recognition, emotion detection, and NLP as ROS components with real-time inference on edge hardware.
  • Owned the ML pipeline from data collection to production inference via TensorFlow Serving and Flask across multiple product lines.

Patents

Co-inventor. "Systems and Methods for Generating a Refined 3D Model Using Radar and Optical Camera Data." US20190295319A1, filed March 14, 2019.

Education

2010 – 2014: Bachelor of Science in Computer Science Honor Programs - University of Science Ho Chi Minh City. GPA: 8.7 / 10

Skills

  • Languages: Python, Rust, C++, TypeScript
  • ML/DL: PyTorch, Multimodal Sensor Fusion, Computer Vision, 3D Reconstruction
  • AI Agents: Multi-Agent Pipelines, Tool-Augmented LLMs, LangGraph, Model Context Protocol (MCP)
  • Frameworks: FastAPI, FastMCP, LangChain, Qt, ESP-IDF, Axum, Tokio, React
  • Infrastructure: Docker, AWS (Lambda, S3, ECS), PostgreSQL, Redis, CI/CD, MLOps

Publications

Book: Machine Learning with TensorFlow 1.x (2017) Co-authored 5 of 13 chapters on Deep Learning with TensorFlow.

Book: OpenCV3 Blueprints (2015) Co-authored 2 chapters: "Recognizing Facial Expressions with Machine Learning" and "Panoramic Image Stitching Application using Android Studio and NDK."

2014: Hong-Quan Hua, T. Hoang Ngan Le, and Bac Le, "An Effective Initialization for ASM-based Methods", CISIM 2014 Conference.