Machine Learning Engineer

Binh Le Do Thanh

Machine Learning Engineer

+848******** Male - 19/10/2000

*********@*****.*** linkedin.com/in/binh234

github.com/binh234 Thu Duc, Ho Chi Minh City,

Viet Nam

Summary

Recent graduate with a Bachelor's degree in Computer Science and a strong interest in artificial intelligence. Skilled in Python, TensorFlow, and PyTorch. Passionate about leveraging AI to solve complex problems and improve user experiences.

HO CHI MINH UNIVERSITY OF TECHNOLOGY 09/2018 - 11/2022 Education

Bachelor of Engineering, Computer Science

Program: Honors program GPA: 8.76/10

SINGALARITY 6/2021 - 9/2021

Work Experience

AI Intern Ho Chi Minh City, Viet Nam

Technologies: Python, Tensorflow, Pytorch, Docker, Spring Framework

- Learned how to use Docker and Kubernetes in the development process

- Conducted experiments in Natural Language Processing and Speech Processing

- Assisted in the development of a number recognition system that achieved an 85% accuracy rate

VIETNAMESE AUTOMATIC SPEECH RECOGNITION 09/2021 - 05/2022 AUTOMATIC CAPITALIZATION AND PUNCTUATION 01/2022 - 04/2022 INVERSE TEXT NORMALIZATION 01/2022 - 03/2022

Personal Projects

Developer Team size: 2

Source:

Description: Develop a fast, accurate, lightweight Vietnamese speech recognition system that can recognize and transcribe spoken words into text using Wav2Vec2.0.

Technologies: Python, Transformers, Speechbrain, Wav2Vec2 Responsibilities: - Collected over 600 hours of raw speech data from multiple sources

- Improved performance by 20% with a customed tokenizer based on Vietnamese acoustic

- Achieved 96% accuracy on the clean test set and 90% accuracy on the noise test set Developer Team size: 1

Source:

Description: The project aimed to predict the correct capitalization and punctuation of text inputs, such as speech recognition output or messages. I used Sequence Tagging approach to tackle this problem. Technologies: Python, Transformers, NLP

Responsibilities: - Preprocessed the data by removing stopwords and performing normalization, tokenization

- Utilized pretrained BERT model for faster convergence and better accuracy

- Achieved 90% accuracy on the Capitalization task and 80% on the Punctuation task Developer Team size: 1

Source:

Description: Develop an inverse text normalization system to improve the accuracy of chatbot responses and speech recognition system.

Technologies: Python, NeMo, NLP

Responsibilities: - Used Finite State Transducer (FST) to map normalized text to its original spoken or written form, considering contextual and linguistic factors

https://github.com/binh234/wav2vec2-vi-asr

https://github.com/binh234/capu

https://github.com/binhnd234/nemo

- Contributed to the NeMo repository and got accepted Programming languages Python, Go, JavaScript, ReactJS, NodeJS, Java, Kotlin, C/C++ Machine learning frameworks TensorFlow, PyTorch, Transformers, SpeechBrain, NeMo, scikit-learn Frameworks/Libraries Django, Express, Fiber, Material UI 5, Bootstrap, Docker Databases MySQL, Firebase, MongoDB

Tools Visual Studio Code, GitHub

Others Data Visualization, Computer Vision, Natural Language Processing, Speech Processing

Skills

AWS Certified Cloud Practitioner Certificate 2022

43rd place in Google Landmark Retrieval 2021 2021

TOEIC Certificate with score 825, issued by IIG 2020 First prize in 2019 Mathematical Olympiad for student 2019

Honors & Awards

Contact this candidate