$ whoami
Trần Quang Huyên
AI Research Scientist & AI EngineerDa Nang, Vietnam
I bridge machine learning research and production systems. I completed an MSc by Research (Distinction, thesis defended) with first-author IEEE papers in quantum ML and medical imaging, and I have spent 4+ years in industry at Everfit and Hekate on R&D and GenAI and computer vision architectures, turning complex models into low-latency, cost-effective products for enterprise and healthcare.
- 4+ yrsindustry AI experience at Everfit and Hekate
- 2first-author IEEE papers in 2025
- 3.7MSc GPA out of 4.0, Distinction
- 22awards and honors since 2016
01
Experience
AI Engineer
Everfit
Sep 2025 – Sep 2026
- AI profile builder & AI interviewer: led end-to-end development of AI-generated seller profiles, from prompt engineering and internal tooling to the voice-interviewer API; ran field-level evaluations, optimized Instagram crawling and fixed critical pre-launch issues for seller onboarding.
- AI nutrition & food recognition: built evaluation and improvement pipelines for production AI, with analytics dashboards, synonym-to-canonical mapping and a multi-tier nutrition cascade; benchmarked 39 VLM configurations, designed the Gold-300 dataset, optimized a compact scan-type classifier and routed A/B tests with Statsig.
- Custom wake-word detection & Voice2JSON: led R&D and the end-to-end PoC for hands-free, voice-to-JSON workout logging, with a custom wake-word pipeline, under 2 s latency at about $0.00036 per request, API migration, conversational logging and the iOS handover.
Research Student
CIS Lab, National Chung Cheng University · Chiayi, Taiwan
Dec 2024 – Mar 2025
- Researched quantum machine learning and quantum self-attention under Prof. Van-Linh Nguyen, implementing classifiers in Python and PennyLane.
- Designed efficient circuit classifiers for quantum self-attention in Vision Transformers, published as a first-author paper at IEEE Globecom Workshops 2025.
- Completed the 2025 International Intern Research Program with an “Excellent” rating.
AI Engineer
Hekate
Nov 2021 – Dec 2024
- R&D on computer vision, NLP, GenAI and LLMs (GPT, Llama, Qwen, Claude) to customer requirements.
- Cerebral artery disease classification: designed, trained and deployed cerebral artery segmentation and disease prediction models in PyTorch (85% accuracy).
- MedNow: built a medical chatbot on SOLAR with RAG over ChromaDB; RAG raised answer accuracy by 20%.
- Milk advertisement violation detection (Alive & Thrive, UNICEF): led a team of 5 engineers covering Lao and Mongolian, reaching 93% accuracy and cutting processing time by 30%.
- Sumi: maintained and optimized an LLM chatbot, raising user engagement by 30%.
02
Research
Peer-reviewed papers, then the student research conference work that led to them.
First author · IEEE
Efficient Circuit Classifier Design for Enhancing Quantum Self-Attention in Vision Transformers
IEEE Globecom Workshops (GC Wkshps), Taipei, Taiwan, pp. 170–175 · 2025
First author · IEEE
Comparative Analysis of Deep Learning Architectures for Brain Tumor Segmentation
International Conference on Advanced Technologies for Communications (ATC), Hanoi, Vietnam · 2025
Student research conference
Research and Enhancement of the Quantum Self-Attention Model
Faculty-level Student Scientific Research Conference · 2025
Third Prize (faculty)
Student Support Chatbot for the University of Science and Technology
Faculty-level Student Scientific Research Conference · 2023
Second Prize (faculty)
Research on the GaitBase Model Applied to Gait-Based Identity Recognition
Student Scientific Research Conference · 2022
First Prize (faculty) · Encouragement Prize (university)
Shoplifting Detection in Retail Stores Using Computer Vision
Student Scientific Research Conference · 2022
03
Projects
Research, product work at Everfit and Hekate, and things I built on my own. Linked ones have public code or a live demo.
Quantum self-attention for Vision Transformers
Circuit-centric quantum classifiers that strengthen quantum self-attention, simulated in PennyLane on downscaled MNIST and the MC language task.
First author · IEEE GC Wkshps 2025
Brain tumor segmentation benchmark
Six U-Net-family models (U-Net, ResUNet, DeepResUNet, DRU, SRU, DUCK-Net) retrained and compared on 3,064 MRI slices; DUCK-Net led with a test Dice of 0.8575.
First author · IEEE ATC 2025
Glaucoma classification from OCT
Master’s thesis: a deep learning model that classifies glaucoma from OCT images of the retinal nerve fiber layer, with a 3D CNN on raw 200³ volumes trained by one PyTorch pipeline from Colab to a multi-node SLURM cluster.
MSc thesis · defended
Gait-based identity recognition
Identifies people by the way they walk using the GaitBase model, wrapped in a Flask web app. Built with a teammate; I also presented the papers we built on.
First Prize (faculty) · Encouragement Prize (university)
Student support chatbot
Answers students’ questions from the university’s own documents through a ChromaDB vector store, after comparing GPT-3, GPT-4, T5, Phi and BLOOM.
Second Prize (faculty)
Shoplifting detection in retail stores
Computer vision that spots shoplifting in store camera footage; a four-person student research project.
AI profile builder & AI interviewer
AI-generated seller profiles for onboarding: prompt engineering, internal tooling and a voice-interviewer API, checked by field-level evaluations and fed by an optimized Instagram crawler.
End-to-end lead, through launch
AI nutrition & food recognition
Evaluation and improvement pipelines for production food recognition: analytics dashboards, synonym-to-canonical mapping, a multi-tier nutrition cascade, a compact scan-type classifier and Statsig A/B routing.
39 VLM configurations benchmarked · Gold-300 dataset
Custom wake-word detection & Voice2JSON
Hands-free workout logging: a custom wake-word pipeline and a voice-to-JSON model, taken from R&D to an end-to-end PoC handed over to iOS.
< 2 s latency · ~$0.00036 per request
Milk advertisement violation detection (Alive & Thrive, UNICEF)
Finds milk ads that break the marketing rules for products for pregnant women and children under 24 months, across Facebook, Instagram and websites, in 5 countries.
Led 5 engineers · 93% accuracy
Cerebral artery disease classification
3D segmentation of cerebral arteries with nnU-Net plus a disease classifier, trained on imbalanced medical scans and delivered as a web platform.
85% accuracy
MedNow medical chatbot
Gives medical guidance with the SOLAR LLM and RAG over WHO documents, and reads uploaded prescriptions with OCR.
+20% accuracy with RAG
Virtual expert
A fine-tuned LLM with embedding search that answers questions about a user’s specialized documents in real time.
Virtual Friend VR game
A virtual-reality companion for Oculus Quest 2, built in Unity with the XR toolkit and pre-trained NLP models.
Virtual QC platform
Transcribes customer–employee calls with speech separation and speech-to-text, then scores sentiment so teams can review service quality.
H.266/VVC compression for 4K CCTV
Studied the H.266 (VVC) standard and machine-learning approaches for compressing 4K CCTV video, tested open-source implementations and built a web demo.
Vietnamese license plate recognition
Fine-tuned YOLOv8 to find Vietnamese plates in CCTV video and read them with OCR, on a dataset I collected.
Store demand forecasting
Time-series models, LSTM among them, that forecast store demand.
PhD search agent
Agent workflows that find PhD positions and labs, score the fit and draft applications, with tested OpenAlex and Bluesky CLIs.
Learning platform for students with disabilities
Lessons, sign-language guides and games for Vietnamese students with speech and language disabilities.
04
Education
2025 – Mar 2027 (expected)
Master of Science (by Research) in Electronics Engineering
The University of Danang – University of Science and Technology
- GPA 3.7 / 4.0 (Distinction).
- Thesis: A Deep Learning Model for Glaucoma Classification Using Optical Coherence Tomography Images of the Retinal Nerve Fiber Layer.
- Supervisor: Prof. Tran Dang Khoa Phan.
- Research and thesis defense completed; awaiting degree conferral.
2020 – 2025
Bachelor of Engineering in Computer Engineering
The University of Danang – University of Science and Technology
- GPA 3.58 / 4.0, 2nd in the major.
- 180 credits, Vietnam Qualifications Framework level 7 (master’s-equivalent).
- Projects: Quantum Self-Attention; replicating the Transformer model.
- Coursework: machine learning, computer vision, data structures and algorithms, electronic circuits, digital logic design, probability and statistics, linear algebra.
05
Awards & honors
2025
- “Excellent” rating, International Intern Research Program, National Chung Cheng University
- Third Prize, Faculty-level Student Scientific Research Conference
2024
- Second Prize, Faculty-level Student Scientific Research Conference
- Outstanding Student in Academics and Activities, Faculty of Electronics and Telecommunication Engineering
2023
- First Prize, Fisico 2023: CYBERA, physics competition for university students
- First Prize, Faculty-level Student Scientific Research Conference
- Encouragement Prize, University-level Student Scientific Research Conference
- Outstanding Student in Academics and Activities, Faculty of Electronics and Telecommunication Engineering
- Faculty commendation for contributions to the Faculty of Electronics and Telecommunication Engineering
- Excellent Student of the academic year
- PIONEER Student Research Club certificate for strong competition results
2022
- Second Prize, ETE Code Contest
- Semi-finalist, AWS Build On Vietnam 2022
- Certificate for the Faculty of Electronics and Telecommunication Engineering booth at BK Techshow 2022
2021
- Excellent Student of the academic year
2020
- Merit-based entrance scholarship, University of Science and Technology
- Second Prize in Physics, Da Nang city-level gifted student exam
2019
- Encouragement Prize in Mathematics, high school gifted student exam
2018
- Encouragement Prize in Physics, high school gifted student exam
2017
- Third Prize in Physics, Da Nang city-level gifted student exam (grade 9)
- Second Prize, district-level online physics contest
2016
- Second Prize in Physics, district-level gifted student exam (grade 8)
06
Teaching & leadership
Aug 2023, Aug 2024
Tutor, Hekate Summer Camp (Seasons 1 & 2)
Hekate Academy
Taught 10–15 students a week deep learning and computer vision (YOLO, R-CNN, OCR) plus machine learning basics.
Mar – Apr 2023
Python Tutor
Faculty of Electronics and Telecommunication Engineering, University of Science and Technology
Tutored 20–25 engineering students a week in Python.
Jan 2022 – Nov 2024
Head of AI Expertise
PIONEER Student Research Club, Faculty of Electronics and Telecommunication Engineering
Trained members in AI and organized competitions; recognized for the club board’s work in 2022–2023.
Aug 2022
Technical Tutor
ROBODNIC 2022, robot contest for Da Nang high school students
Guided a team through Arduino, mechanical and electrical design.
Oct 2021 – Aug 2022
Head of Expertise
STEAM for Youth (S4Y)
Owned the quality of all STEAM content.
Jan 2019 – May 2021
Founder & President
HHT Coder and Scientist, Hoang Hoa Tham High School
Started and ran the school’s science and technology club.
07
Skills & certifications
AI & machine learning
Research domains
Programming
MLOps & cloud
Data & backend
Certifications
- TOEIC 880
- AI for Medical Diagnosis (DeepLearning.AI)
- Custom Models, Layers, and Loss Functions with TensorFlow
- Python for Data Science, AI & Development
- Introduction to TensorFlow for Artificial Intelligence, Machine Learning, and Deep Learning
08
Get in touch
Questions, collaborations or opportunities: email is the fastest way to reach me.