Dinh Khai Nguyen

Basé au Vietnam, je suis chercheur et ingénieur en intelligence artificielle, passionné par les technologies guidées par les données. Mon parcours académique en informatique et mon expérience pratique en apprentissage automatique, deep learning, traitement du langage naturel et fintech m’ont permis d’acquérir une solide rigueur théorique et opérationnelle. J’ai mené des projets variés allant de la vision par ordinateur et la détection de fraudes aux agents conversationnels et systèmes temps réel. Mon objectif est de bâtir des solutions d’IA créant un impact mesurable et durable.

Apprentissage automatique

PyTorch · Transformer · Seq2Seq · YOLOv5 / YOLOv1 · CBAM · XGBoost · SMOTE / ADASYN · scikit-learn

Traitement du langage naturel

Vietnamese NLP · Diacritic restoration · BLEU / ChrF++ · Sentiment classification

Systèmes & Déploiement

FastAPI · Docker · PostgreSQL · React · Apache Spark · Apache Hadoop

Inférence & Performance

TensorRT · ONNX Runtime · TorchScript · torch.compile · FP16 / INT8 · CUDA profiling

• ResearchMap AI2026

  Turns a one-line description of a research field into a map of its literature: topics, trends and gaps. An agent answers questions about the field, citing only papers it actually retrieved.

  Cited papers verified real 36/36 · Retrieval precision@5 0.867 · Cost per answer $0.00032

• TaxMate AI2026

  A tax-compliance assistant for Vietnamese online sellers. It answers legal questions with a citation on every claim and reconciles sales against declared revenue. Built by a team of four that Khai led.

  Legal recall@5 93% · Legal documents indexed 13 · Cores integrated 4

• Detoura2026

  A realtime two-player board game built as a platform. It has a deterministic rules core, one set of typed schemas that validates, documents and tests the API, and a separate tournament service for bots.

  Workspace packages 9 · Interfaces under contract 5 · Protocol messages 14

• Bacera2026

  An online shop and workshop booking site for a Hanoi pottery studio, built on WordPress with Pancake POS as the system of record. Its public demo replays recorded POS traffic.

  Integration modules 6 · Custom SQL tables 15 · POS endpoints replayed 12

• Flow Center2026

  Booking, a self-assessment and a chat assistant for a clinic that treats period pain with traditional medicine. The AI stays strictly inside clinician-approved content.

  Parallel bookings in the race test 8 · Warning signs that stop the analysis 10 · Days a conversation is kept 90

• AI Pipeline Mastery2026

  A course built as software. A Python pipeline generates the course material and executes it as a test, and a Next.js platform runs learners' code either in the browser or on a pool of remote GPUs.

  Slides generated 383 · Figures rendered 129 · Execution lanes 2

• Classroom Trash Detection2026

  A computer-vision system split across a network. A stateful client handles the camera and the dashboard, and a stateless FastAPI service runs YOLOv8 on a GPU.

  Model formats 3 · Upload width 640 px · Deploy targets 3

• Neural Machine Translation2025

  Built a 12.5M-pair dataset for Vietnamese diacritic restoration and implemented RNN, GRU, LSTM and Transformer seq2seq models from scratch in PyTorch.

  BLEU 81.31 · ChrF++ 88.57 · Sentence pairs 12.5M

• Optimization for AI Inference Engines on GPUs2025

  Optimising GPU inference for ResNet50 and DistilBERT with FP16, INT8, TorchScript/torch.compile and ONNX.

  FP16 throughput ≈4× · INT8 speedup 2–5× · Graph optimisation 1.2–1.8×

• License Plate Recognition System2025

  A real-time license plate recognition system: a fine-tuned YOLOv5 model with OCR, a FastAPI and PostgreSQL backend and a React frontend, deployed with Docker.

  Throughput ~12 FPS

• Quantitative Trading System2025

  A quantitative trading system built from scratch in PyTorch: a Transformer that predicts BUY/SELL/HOLD signals from financial indicators and historical trading data.

  Validation accuracy 92.75% · Test accuracy 92.50% · Training samples 3.4M

• Credit Card Fraud Detection2024

  Fraud detection on the Kaggle credit card dataset, where 0.172% of transactions are fraud, comparing four classifiers with SMOTE and ADASYN resampling.

  ROC-AUC 0.9817 · Recall 0.816 · Positive rate 0.172%

• Large-Scale Text Sentiment Classification2024

  Sentiment classification of 1.6M tweets (Sentiment140) with Naive Bayes and SVM, run on Apache Hadoop and Apache Spark.

  Dataset 1.6M · SVM accuracy 80–82% · Naive Bayes accuracy 75%

• YOLOv1 With CBAM From Scratch2024

  YOLOv1 rebuilt from scratch in PyTorch, with its own dataset class and loss function, and CBAM attention added to the feature extractor.

  Train mAP 0.8829 · Best val mAP 0.6994 · Parameters 269.2M

• Troisième Prix, Prix de recherche scientifique étudiante — Intelligence Artificielle — VNU University of Engineering and Technology2026
• Troisième Prix, Volet Étudiant — Hackathon Logistique Humanitaire — Humanitarian Logistics Hackathon2026
• Bourse d’encouragement aux études universitaires — University of Engineering and Technology2024
• Deuxième Prix, Concours provincial d’excellence scolaire en Physique — Thai Nguyen Province2023
• Troisième Prix, Concours provincial d’excellence scolaire en Informatique — Thai Nguyen Province2021