Gi-Luen Huang

(Allen)

Senior AI Engineer & Assistant Project Manager · CyberLink
Download CV See experience Get in touch
01

About

Gi-Luen (Allen) Huang

I'm a Senior AI Engineer & Assistant Project Manager at CyberLink, specializing in computer vision, deep learning, and generative AI. My first stretch here was spent deep in Flux — training LoRA for object removal & replacement and product backgrounds, then Flux Kontext for human beauty effects.

Currently I guide team members building GenAI features and drive our adoption of agentic development. Before CyberLink, I shipped automotive AI at MobileDrive — projecting vehicle trajectories from front-view cameras, then quantizing and deploying the model on the Qualcomm 8295 platform for on-vehicle testing.

I earned my M.S. in Data Science from National Taiwan University (GICE) in 2023, advised by Prof. Pei-Yuan Wu, with 4+ years of research across deep learning, computer vision, and machine learning.

  • LocationNew Taipei, Taiwan
  • FocusGenerative AI · Computer Vision
  • DegreeM.S., NTU — GPA 4.30, rank 1/160
  • PublicationsICIP · IEEE Access · ICSSE
02

Experience

CyberLink
Aug 2024 — Present
New Taipei, Taiwan

Senior AI Engineer & Assistant Project Manager

May 2026 — Present
  • Guide team members in developing GenAI-powered features, and help plan feature roadmaps and release timelines.
  • Established and shared best practices for coding-agent adoption, cutting unexpected outputs and improving development efficiency.
  • Designed and built an internal Agentic AI PoC platform integrated with company auth, letting PMs and designers accelerate their product and design workflows.

Senior AI Engineer

May 2025 — May 2026
  • Trained LoRA on Flux for adaptive product background generation and object removal.
  • Trained LoRA on Flux Kontext for a human beauty-effect feature that outperformed Meitu in internal QA.
  • Built an image-to-video feature based on Wan2.2-Animate.

AI Engineer

Aug 2024 — May 2025
  • Developed Flux-based features trained with LoRA and ControlNet — text-to-image and product background generation.
  • Built and shipped generative fill — the #1 most-used feature in PhotoDirector Mobile, beating Adobe Firefly, SnapEdit & TouchRetouch in internal QA.
MobileDrive Technology
Jun 2023 — May 2024
New Taipei, Taiwan

Machine Learning Engineer

  • Projected future vehicle trajectories onto front-view images as ground truth for a 3D trajectory model, training with minimal labeling from RTK + camera data.
  • Developed and optimized the 3D vehicle trajectory model; quantized it for embedded deployment.
  • Deployed on Qualcomm 8295 (Python → C++) at 200 FPS, cutting L1 error from 4.0 to 0.15.
  • Integrated the model into the AR navigation algorithm for on-vehicle testing, improving hard scenarios like roundabouts and sharp turns.
Jubo Health
Jul 2022 — Aug 2022
New Taipei, Taiwan

Machine Learning Engineer Intern

  • Built general frameworks for recognition, segmentation, and object detection.
  • Enhanced the wound-classification model, improving accuracy ~3%; deployed models via Docker on GCP.
Neurobit Technologies
Feb 2022 — Jun 2022
Taipei, Taiwan

Machine Learning Engineer Intern

  • Built a self-supervised gaze-estimation model that reduced error rate by 90%.
  • Designed a feature-matching algorithm to detect torsional eye rotation.
TSMC
Jul 2021 — Aug 2021
Hsinchu, Taiwan

Information Technology (IT) Intern

  • Full-stack system integration.
Teaching Assistant
  • Deep Learning for Computer Vision — NTU, 2022 (Prof. Yu-Chiang Frank Wang)
  • Machine Learning — ITRI, 2022 (Prof. Pei-Yuan Wu)
  • Time-Frequency Analysis & Wavelet Transform — NTU, 2021 (Prof. Jian-Jiun Ding)
  • Data Structures — NTU, 2021 (Prof. Pei-Yuan Wu)
03

Selected Work

CTGAN
Generative · ICIP 2022

CTGAN

Cloud Transformer Generative Adversarial Network — published at IEEE ICIP 2022.

View on GitHub ↗
STSANet
Saliency · Video

STSANet

PyTorch implementation of the Spatio-Temporal Self-Attention Network for video saliency.

View on GitHub ↗
STAS Segmentation
Segmentation · Top 1%

STAS Segmentation

Lung adenocarcinoma pathology segmentation — 2/307 on the T-Brain private leaderboard.

View on GitHub ↗
Crop Status Monitoring
Classification · Top 1%

Crop Status Monitoring

Crop-status recognition by image classification — 3/428 on the AIdea private leaderboard.

View on GitHub ↗
03.1

Publications

IEEE Access2023
Viewing Bias Matters in 360° Videos Visual Saliency Prediction.
P.-W. Chen, T.-S. Yang, G.-L. Huang, et al.
IEEE ICIP2022
CTGAN: Cloud Transformer Generative Adversarial Network. M.S. Thesis
Gi-Luen Huang, Pei-Yuan Wu
ICSSE2020
Face Expression and Tone of Voice for Deception System. Undergrad Research ★ Best Student Paper
L.-W. Hsiao, J.-M. Guo, G.-L. Huang, et al.
04

Skills

Generative AI

Diffusion Models Flux Flux Kontext LoRA ControlNet Image-to-Video (Wan2.2)

Deep Learning

PyTorch Model Quantization Edge Deployment (Qualcomm 8295) 3D Computer Vision

Languages

Python C++

Tools & Agentic

Git Docker AI Coding Agents (Claude Code · Cursor)

05 — Contact

Let's build something.

Open to conversations about generative AI, computer vision, and shipping ML to production. The fastest way to reach me is email.

giluen.huang@gmail.com