Skip to content
View trancongman276's full-sized avatar
👉
Moody as always 🐧
👉
Moody as always 🐧

Highlights

  • Pro

Block or report trancongman276

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
trancongman276/README.md

Hi there, I'm Mẫn 👋

AI / Machine Learning Engineer & Computer Vision Specialist

Specializing in Efficient Deep Learning, Computer Vision Pipelines, Generative Architectures, and Edge Optimization.

GitHub Gmail


💡 About Me

  • 🔬 Currently conducting cutting-edge research on Generative AI & State Space Models (SSMs) for 3D trajectory sequence recognition.
  • ⚡ Experienced in building end-to-end Computer Vision pipelines and optimizing deep learning inference engines for edge/production deployment.
  • 🏗️ Skilled in designing modular, decoupled AI system architectures with high hardware cost efficiency.
  • 💬 Ask me about PyTorch, Mamba-v2, TensorRT, OpenCV, and Edge Optimization.

🔬 High-Impact Research

📄 Generative AI for 3D Sequence Recognition (Paper in Progress)

Designing the world's first generative architecture tailored for continuous 3D trajectory synthesis and sequence recognition.

  • Novel Architecture: Designed an Encoder-Decoder generative framework integrating Temporal Convolutional Networks (TCN) and Mamba-v2 (SSMs) to capture complex spatial-temporal patterns.
  • Benchmark Dataset: Curated a pioneer trajectory dataset featuring 20,000+ continuous samples, collected from 10+ participants across ~2,000 fine-grained character and word labels.
  • Performance: Outperformed traditional sequential models (RNNs/Transformers) in long-sequence spatial modeling while significantly reducing computational complexity and inference latency.

🛠️ Key Engineering Projects

Project Real-Time Edge Vision & Analytics Platform
  • Role: Computer Vision & Optimization Engineer
  • Impact: Optimized deep learning inference pipelines using YOLO, ByteTrack, TensorRT, and ONNX, achieving a ~50% throughput increase (FPS) on hardware-constrained edge nodes.
  • Key Features: Multi-camera object detection, vehicle tracking, Automatic Number Plate Recognition (ANPR), and real-time speed estimation using OpenCV.
  • Tech Stack: PyTorch OpenCV TensorRT ONNX ByteTrack YOLO
🏗️ Project Modular Smart Analytics System Architecture
  • Role: Lead AI Systems Architect
  • Impact: Architected a highly adaptable, decoupled AI platform featuring plug-and-play inference modules and asynchronous queues, supporting flexible multi-tier hardware deployments.
  • Key Features: Comparative benchmark evaluations across SOTA vision models to balance real-world accuracy and cloud computing overhead.
  • Tech Stack: System Architecture Python REST APIs Modular Design Docker
🖼️ Project Vision-LLM Spatial Quantification Engine
  • Role: AI Solutions Researcher
  • Impact: Automated complex blueprint data extraction by combining multimodal LLMs with classical spatial algorithms.
  • Key Features: Utilized multimodal APIs (Gemini AI) and advanced prompt engineering to clean input blueprint noise, paired with server-side lightweight OCR and OpenCV geometric routines for automated floor area calculations.
  • Tech Stack: Vision LLMs Prompt Engineering OpenCV OCR Geometric Algorithms
📱 Project Cross-Platform Inference & Dataset Curation
  • Role: ML & Deployment Engineer
  • Impact: System performance auditing and cross-platform compilation feasibility assessments.
  • Key Features: Profiling ONNX Runtime on desktop environments, surveying native iOS deployment viability (CoreML compilation), and curating high-precision datasets for YOLOv9-DEYO models.
  • Tech Stack: ONNX Runtime CoreML Feasibility Data Engineering YOLOv9

🧰 Tech Stack & Tools

Category Technologies / Frameworks
AI / Deep Learning PyTorch Mamba-v2 TCN Vision LLMs Prompt Engineering YOLOv9-DEYO
Computer Vision OpenCV ByteTrack DeepSORT Object Tracking Segmentation ANPR
Inference & Optimization TensorRT ONNX Runtime Model Quantization Edge Computing C++ / Python
Systems & Tools Modular Architecture REST APIs Docker Git Linux

📊 GitHub Stats

Mẫn's GitHub Stats
Top Langs

Pinned Loading

  1. yolov7-pose yolov7-pose Public

    Pose Detection based on YoloV7 deployed on ONNXRuntime

    Jupyter Notebook 24 4