Terrance Duc - Senior Al Engineer

Accomplished Data Scientist/AI Engineer with more than 6 years of experience in extracting, analyzing and interpreting large amounts of data from a range of sources, using algorithmic, data mining, artificial intelligence, machine learning and statistical tools, to make it accessible to businesses. Having a big passion for delivering valuable data through analytical functions and data retrieval methods. Adept at receiving and monitoring data from multiple data streams. Bringing forth the ability to synthesize quantitative information and interact effectively with colleagues and clients.

Project list

Intelligent Retail Automation

Develop and implement a cutting-edge computer vision software solution for retail, specifically targeting the grocery industry. The system will utilize Al-powered product identification to minimize checkout errors, reduce plastic packaging waste, and enhance operational efficiency. By streamlining the shopping experience, this project aims to deliver a faster, more convenient, and sustainable retail experience for customers while driving business growth for grocers.

- Develop and train generative Al models using frameworks such as TensorFlow, PyTorch - Implement various NLP tasks, such as text classification, sentiment analysis, named entity recognition, and language translation, using generative Al models - Continuously monitor and optimize generative Al model performance to ensure high accuracy, efficiency, and scalability - Develop and implement data pipelines to collect, process, and analyze data from various sources

Technologies: Python, TensorFlow, PyTorch, Google Cloud Platform (GCP)

Flyer

FLYER is the first high quality virtual interactive english testing platform for kids.

- Study existing database, data flow, business - Clean and transform, manipulating data for analysis and visualization - Create Google Data Studio dashboards for mayor performance metrics, based on customer expectation - Design, Develop and Optimize Automatic Scoring for Writing Tests - Design, Develop and Optimize Automatic Scoring for Speaking Tests - Design, Develop and Optimize Image, Quizzes Automatic Generation to use as learning contents.

Technologies: Azure, PostgreSQL, MongoDB, Python, TensorFlow, Scikit-learn, PyTorch, Machine Learning, Deep Learning

Healthcare Platform

The client is one of the largest healthcare services groups in Asia. They provide world-class distribution, commercial solutions, patient care, clinic reach, data analytics, pharmacy channel and payor services to support the growing healthcare needs in this region.

- Study existing database, data flow, business - Clean and transform, manipulating data for analysis and visualization - Design, Develop and Optimize ML Classification Model for group merchants categories - Design, Develop and Optimize Automatic Scoring for Writing Tests - Design, Develop and Optimize ML Recommendation System for Inventory optimization.

Technologies: PostgreSQL, Azure

INVETO

Online data analysis platform with Al for investors. Products on INVETO platform include cryptocurrencies, securities, real estate. The system integrates algorithms from Al and ML to help suggest buying and selling products, giving information, trends and price history of products.

- Study existing database, data flow, business - Clean and transform, manipulating data for analysis and visualization - Apply "big data" solutions to solve current business problems - Design, Develop and Optimize ML Recommendation System for buy/sell decision - Design, Develop and Optimize ML Stock New Analyst and summary using ChatGPT

Technologies: Node.js, PostgreSQL, TensorFlow, AWS, Kubernetes, Docker, Scikit-learn, Python, PyTorch, Machine Learning, Deep Learning

AlSuviet

AlSuViet acts as a wise advisor to help users easily access historical knowledge from reliable sources through chatting with the chatbot. At the same time, through digitizing documents, the platform also contributes to storing and preserving knowledge for future generations.

- Study existing database, data flow, business - Clean and transform, manipulating data for analysis and visualization - Design, Develop and Optimize Al Chatbot using local knowledge using Google PaLM v2 - Analyze and Implement Training Algorithm - Design, Develop and Optimize ML Classification Model to detect violated message, questions

Technologies: AWS, PostgreSQL, Python, TensorFlow, PyTorch, Scikit-learn, Machine Learning

Gaspol

The one and only Automotive - Sport & Digital Media Platform for your daily mobility needs

- Study existing database, data flow - Clean & transform data for analytics - Create Google Data Studio/Superset dashboards for mayor performance metrics, based on customer expectation - Migrate data from legacy system to new one - Fix various data bug

Technologies: PostgreSQL, Java

VECA

A Startup in Vietnam who provides a matching system for users and dealers in collecting rubbish.

- Study existing database, data flow - Clean & transform data for analytics - Create Google Data Studio view for mayor performance metrics, based on customer expectation

Technologies: MySQL

Confidential Client

The project, entitled Statistical analysis of mechanical measurements during a full-scale test of reinforced asphalt pavement, allows a better understanding of the full-scale test results, and consequently helps evaluate pavement performances.

- Statistical analysis of mechanical measurements during a full-scale test of reinforced asphalt pavement. (Python) - Collect data and design a dataset of important characteristics of a pavement structure. - Study and interpret that large dataset using Panel analysis, while following Cross Industry Standard Process for Data Mining (CRISP-DM) methodology

Technologies: PostgreSQL, Python, Azure

Confidential Client

- Predict duration of taxi trips in New York City. (Java, R) - Solve the data mining problem, using models (linear regression and artificial neural network), while following CRISP-DM methodology.

Technologies: PostgreSQL, Java

Confidential Client

There are many users in a social network, most of them have missing info (attributes). The project helps to find a way to complete those profiles and detect influencers (users with a lot of connections).

- The illusion of hidden personal data. (Python) - Fill incomplete profiles and find influencers in a network, using community detection methods (hierarchical clustering, Louvain method and simulated annealing), while following CRISP-DM methodology

Technologies: Azure, PostgreSQL, Python

Project list

Education

Techstacks