Ho Chi Minh

DINH CONG HUY HOANG

Ho Chi Minh city, Vietnam

+*4 (902) 891-203 ********************@*****.*** linkedin/dchhoang github/QuiQuang EDUCATION

University of Science-VNUHCM, Ho Chi Minh city, Vietnam October 2021 Present Bachelor of Science in Computer Science

ACADEMIC EXPERIENCE

Gradients Technologies Ho Chi Minh city, Vietnam

AI Engineer Intern April 2024 October 2024

• Worked at an outsourcing company specializing in computer vision, focusing on camera and digital image processing.

• Led data processing efforts, customizing and training models to meet client-specific requirements.

• Evaluated project results and developed strategies for future improvements.

• Contributed to research and publications, advancing technical knowledge in the field of computer vision. PROJECTS

GRAPH EHR: October 2024 Now

About:

• This project develops CL-ETM, a framework for learning robust representations and predicting diseases from Electronic Health Records (EHRs). It addresses the complexity of EHR data using advanced techniques:

– Dynamic Hypergraph Modeling: Captures temporal relationships in medical events.

– Transformer Architecture: Integrates visit-level and temporal information.

– Contrastive Pretraining: Combines diverse modalities (e.g., labs, diagnoses).

– Hierarchical Attention: Prioritizes key medical events for interpretability.

– Meta-learning: Enables few-shot prediction of rare diseases. Team member: 2

Responsibilities:

• Optimized parallel processing to enhance the efficiency of EHR data handling.

• Implemented and trained the contrastive learning framework for robust multimodal patient representations.

• Analyzed datasets, performed statistical evaluations, and proposed solutions to improve model performance. INPAINT CONTEXT: Github September 2024 October 2024 About:

• Constructed a model for object removal in images.

• Modified the training format and adjusted the masking approach of the MAE model to align with specific objectives. Team member: 3

Responsibilities:

• Co-led the development of a model for object removal, handling planning, researching, implementation, and training.

• Collected, processed, and prepared a custom dataset focused on furniture for model training.

• Contributed to the model’s conditioning layer design and fine-tuning for optimal performance in object removal tasks. VIDEO RETRIEVAL: Github August 2024 September 2024 About:

• Developed a system to extract information from videos using language models and vision-language models.

• Implemented quantization techniques to enhance computational efficiency and improve retrieval accuracy.

• Stored extracted data in a vector database to ensure optimized access and scalability.

• Focused on leveraging advanced models for precise and efficient information retrieval from video content. Team member: 4

Responsibilities:

• Implemented data storage using the vector database Qdrant and MySQLite, integrating Qwen and Jina models to generate embeddings for efficient information extraction. Dinh Cong Huy Hoang July 2024

YOLO MULTITASKING May 2024 September 2024

About:

• Development and customization of a YOLO model for specialized tasks such as object detection and human gaze recognition.

Team size: 6 members

Responsibilities:

• Customizing the YOLO architecture to perform specific tasks like object detection and human gaze recognition.

• Developing optimized workflows to accelerate the training process. ONNX RUNTIME: Github June 2024 August 2024

About:

• Developed a library platform to execute ONNX models, facilitating seamless integration between Python and C++.

• Ensured compatibility and optimal performance across different hardware and software environments, promoting efficient execution of machine learning models.

Team member: 2

Responsibilities:

• Utilized ONNX Runtime for efficient model management and execution across various hardware providers (CPU, GPU, NPU).

• Created a user-friendly Python API to load, configure, and run models, while the C++ backend optimized performance and resource management.

• Establish bindings between Python and C++ for seamless integration. FACIAL TRACKING: Github January 2024 April 2024

About:

• Developed a deep learning-based system to prevent cheating in online exams.

• Implemented real-time detection of real vs. spoof faces using the MobileNet model.

• Integrated MediaPipe for extracting facial landmarks to monitor user attention.

• Designed a user-friendly desktop application with PyQt5 for seamless interaction.

• Ensured robust performance for real-time usage scenarios. Team member: 4

Responsibilities:

• Collected and processed image data for training and evaluation.

• Led the project as the main developer, overseeing core implementation and processing. SKILLS

• Technical: Python, Julia, Java, C/C++, CUDA Programming, ...

• Databse: mySQL, PostgreSQL, mongoDB, ...

• Libraries: PyTorch, TensorFlow, NumPy, Pandas, ...

• Software: Pyspark, Hadoop, Docker, Git, Tableau, ...

• Soft Skills: Problem Solving, Teamwork, Hard-working, Responsibility, Communication, ...

• Language: English, Vietnamese

CERTIFICATIONS

• IELTS: Overall 6.5 (December 2020)

2

Contact this candidate