Mandy Nguyen

4.9

Select me!

Mandy is available for hire

Hire Now

Mandy Nguyen

4.9

Gender

Male

Language

English - Advanced

Main expertise

Data Analysts Data Scientists Machine Learning AI Deep Learning Python TensorFlow NLP PyTorch Data Visualization Recommendation Systems Image Processing Software

Introduction

Mandy Nguyen is a seasoned data scientist with six years of experience working at BAEMIN Vietnam and Zalo. She is also the founder of Emolyzer, an AI-driven startup. Her entrepreneurial journey has honed her leadership, team management, and problem-solving abilities. Mandy is skilled in financial analytics, as evidenced by her published research, and has developed machine learning pipelines for tasks such as image classification, text analysis, and user behavior modeling. She collaborates effectively with cross-functional teams to deliver data-driven solutions that enhance customer satisfaction, operational efficiency, and revenue growth. As a Senior Data Scientist, she has successfully implemented machine learning solutions for various data products using Python, TensorFlow, and PyTorch. Mandy is eager to apply for the Senior Data Scientist position to contribute to the development of data solutions that enhance user experiences.

Project list

Emolyzer

2024 - 2024

Emolyzer is an AI-Powered Experience Management Platform focused on real-time analysis of customer feedback. It provides actionable insights to improve customer experiences.

Built machine learning and NLP models for real-time sentiment analysis and predictive insights from customer feedback. Led the integration of AI models into the platform, ensuring performance, scalability, and user-friendly visualizations. Managed and mentored a team of data scientists, providing technical guidance and fostering collaboration. Defined the product vision and translated market needs into product features. Secured funding through Antler VN6 Cohort and engaged with investors to support continued innovation. Worked with clients to customize solutions and ensure Emolyzer's competitive advantage.

Technologies: Machine Learning, NLP

Comprehensive Food Taxonomy and Classification System - Vietnam

2021 - 2023

The initial food categories in the application were inconsistent and incomplete, failing to align with customer needs and preferences. This project focused on developing a robust food taxonomy that meets business objectives and establishing a classification system capable of categorizing dishes rapidly and accurately.

Collaborate with the business team to create a food taxonomy system that aligns with business goals. Build a rule-based NLP classifier to categorize food items into 86 categories with an average accuracy of 95%. Develop a pipeline to automatically classify ordered dishes and user search keywords daily, providing insights into user and market behavior. Visualize search trends by hour and geographical location to support supply-demand analysis, utilizing tools like GeoPandas and Google Data Studio.

Technologies: NLP, Machine Learning, Data Pipeline, Visualization, Geographical Analytics

BAEMIN Billing Image Validation System - Vietnam

2021 - 2023

The project focuses on solving the issue of invalid billing images submitted by riders of BAEMIN, a food delivery service. These images either lack a bill or a valid bill ID, leading to delays and errors in the payment processing. The manual verification process for these billing images has resulted in a high workload for staff, necessitating a more efficient and accurate system for detecting invalid images.

Improved the system to detect invalid billing images, specifically those without a bill or with a bill that lacks a valid ID. Enhanced system efficiency, reducing the number of images requiring manual checks by 50%, while maintaining the detection rate of invalid images.

Technologies: Machine Learning, OCR

Merchant Embedding Without Attributes - Vietnam

2021 - 2023

This project focuses on developing a merchant recommendation system for BAEMIN users. It aims to enhance the user browsing experience by suggesting merchants similar to those recently viewed by the user, making it more convenient and efficient for users who explore multiple merchants before making a decision.

Create a vector representation of merchants based on user in-app behavior. Compare merchant similarity using vector distance. Use merchant vectors for recommendation purposes.

Technologies: Metric Learning, Recommendation System

Harmful Content Classification for Images and Videos

2018 - 2021

The goal of this project is to develop an AI model that can automatically classify images as harmful or normal with high accuracy for Zalo, the largest social network in Vietnam. With over ten million public images uploaded daily, manual labeling is not feasible, making this AI solution essential for managing harmful content.

End-to-end training and deployment of a machine learning model using the TensorFlow framework to predict if images or videos contain sensitive content, achieving an accuracy of 99%. Applied active learning techniques to reduce labeling costs and enhance the model's adaptability.

Technologies: TensorFlow, Active Learning, Redis

Vietnamese Language Normalization for Press Content

2018 - 2021

The project focuses on developing a text normalizer that converts Vietnamese articles into plain text suitable for speech synthesis. It addresses challenges posed by symbols, numbers, abbreviations, and other non-standard text forms within articles, ensuring they are transformed into a format appropriate for speech synthesis. This effort is a component of a broader initiative to create an AI system capable of generating speech from diverse article types. Further details can be found at: https://zalo.ai/experiments/text-to-audio-converter.

Building an optimal Vietnamese normalization system for reading press content in real-time for all articles on https://baomoi.com/. The system automatically corrects common errors encountered by users when entering data. Ensuring accurate reading of over 95% of popular Vietnamese press content.

Technologies: NLP Rules building

Project list

Education

Expertise