AI Engineer
Resume:
HCMC University of Technology and Education **** - 2022
• Major: Mechatronics Engineering Technology
• Degree grade: Distinction
• GPA: 8.26/10 (3.28/4)
• Graduate thesis: Design and manufacture a Person Following Robot (smart cart) apply in supermarkets or any indoor environments. (3 members)
o Description: The robot can follow a person on various complex floors in indoor environments under many challengings: illuminating conditions, pose changes, occlusions, transiently losing the target, the target being covered, etc...
o Technologies: Convolutional neural network, Object detection, Object tracking, Vision transformer; 3D vision, point cloud processing; ROS
o Responsibilities:
Researched and implemented a lightweight model for the person detection problem on an edge device
Working with the Realsense camera, processed various types of input data (2D and 3D) as a signal for locating people and robot navigation
Researched and experimented State-of-the-art Single object tracking (Vision transformer) models for the person-tracking problem in complex environments
Solved the speed problem in the whole system of the robot on Jetson Nano to approximately 10 FPS
Designed and developed ROS system to make the robot steadily operate and user-friendly o Link Github: https://github.com/anhbantre/PersonFollowingRobot Nguyen Huy An
EDUCATION
30/10/2000
**************@*****.***
Ho Chi Minh, Viet Nam
https://www.linkedin.com/in/an-nguyen-huy/
Phan Boi Chau High School 2015 - 2018
• Degree: Good
• Academic awards:
o Won the Third Prize, Mathematics in the Provincial Competition for excellent students in grade 12 2017 - 2018 o Won the Third Prize in Provincial Handheld Computer (Casio) Math Competition 2016 - 2017 o Bronze Mathematics Medal in The National 30/4 Ho Chi Minh City Olympic Competition 2015 - 2016 PROFESSIONAL EXPERIENCE
Realtime Robotics, Viet Nam
AI Engineer 09/2021 - present
Project: Smart playload (5 members)
• Description: The powerful system with the potential to exploit and provide aircraft and onboard payloads with outstanding intelligent features
• Technologies: DeepStream, Object detection, Object tracking, ROS, Socket, Git
• Resposibilities:
o Designed the DeepStream pipeline that is able for real-time analytics on video data and dynamically changes elements to switch between detection and tracking mode (multi-model in one pipeline) o Designed and managed the ROS architecture for the whole system on Jetson Xavier NX. Defined data structure convention of ROS (internal system) and Socket (to communicate with the ground control station) o Reviewed code and tested features to ensure their quality. Set up Gitflow convention and wrote documentation to help other developers faster and easier
o Working direcly with UI/UX team and CEO to generate the product roadmap Project: Powerline Inspection (2 members)
• Desciption: Use drone to detect and observe the components of the electric pole automatically
• Technologies: Object detection, Pytorch, TensorRT, CUDA
• Responsibilities:
o Accelerated the detector’s inference speed approximately from around 60 to 200 FPS and reduced a haft of the running time of the whole system on RTX 3060 GPU
o Improved the accuracy of model utilizing data augmentation o Developed the architecture to implement a multi-model in an end-to-end system
• Programming Languages: Python, C/C++, Matlab
• Frameworks: Pytorch, Tensenflow/Keras, OpenCV, Pandas, Open3D, ROS, etc
• Cloud Service Platforms: Google Colab and Kaggle
• Database: MySQL
• Hardwares: Jetsons (Nano, Xavier NX), Raspberry Pi
• English: Able to read materials, teamwork communication
• Others:
o Experience in implementing baseline AI models to solve practical problems o Experience in working on Linux system, using Git, Docker Other Projects:
• Surveyed, tested and evaluated Vision Transformer models to apply on Jetson Nano
• Working on Lidar driver for streaming real-time videos
• Built a lightweight model for identifying collision situations and the steering angle of the drone at around 120 FPS
• Built a classification model achieves 95% accuracy by applied fine-tuning, hyperparameters optimization, layer freezing, and built a recognition model using One Shot Learning AI Research Intern 07/2021 - 09/2021
• Image processing for practical problems
• Applied re-trained models to detect objects and evaluated model performance using metrics such as accuracy, precision, recall and mAP by 11-point interpolation method
• Wrote programs to control drone in simulator enviroments
• Applied Deep Learning on an edge device for drone navigation in complex simulator environments TECHNICAL SKILLS
CERTIFICATES
• Deep Learning Specialization - DeepLearning.AI, Coursera 02/2023 o Neural Networks and Deep Learning
o Improving Deep Neural Networks: Hyperparameter Tuning, Regularization and Optimization o Structuring Machine Learning Projects
o Convolutional Neural Networks
o Sequence Models
• AI For Everyone - DeepLearning.AI, Coursera 03/2021
• Convolutional Neural Networks in TensorFlow - DeepLearning.AI, Coursera 02/2021
Contact this candidate