OBJECT DETECTION PROJECT – WEST ASIA
AI Developer, May 2025 – Now
Project description
- Working on a team of 5 members including 1 Project Manager, 3 Developers and 1 QC.
- The project is designed to detect objects in images streamed from a large network of cameras. In addition, we built a Label Studio interface to annotate data and continuously improve the model’s accuracy
Responsibilities
- Train and fine-tune the model on a custom dataset covering over 100 object classes
- Implement a Kafka consumer pipeline to ingest concurrent AWS S3 image URLs and queue them for processing
- Integrate with the Triton inference server to perform low-latency, real-time object detection
- Persist detection outputs (bounding boxes, class labels, confidence scores) into MongoDB for downstream analysis
- Build and maintain a Label Studio workspace for annotating new data and driving iterative model improvements
- Develop and deploy a scalable Flask application
Technologies
DEVICE ISSUE HANDLER AGENT - MALAYSIA
AI Developer, April 2025 – May 2025
Project description
- Working on a team of 4 members including 1 Project Manager, 2 Developers and 1 QC.
- The agent is designed to generate relevant test cases and troubleshooting steps for a device issue. Therefore, the engineer will save a lot of time fixing the device.
Responsibilities
- Ingest data from manual document file into Supabase vector store.
- Develop Agent connecting with tools to retrieve device methodology for generating testcases related to the customer defects and critical parameters
- Receive results of test cases and save them in Postgres SQL for tracing issues.
- Develop and deploy scalable FastAPI applications for logging data. Deploy N8N workflow
Technologies
- Python, FastAPI Framework
WEBSITE KNOWLEDGE AGENT PROJECT - CANADA
AI Developer, January 2025 – April 2024
Project description
- Working on a team of 6 members including 1 Project Manager, 3 Developers and 2 QC.
- The agent is designed to scrape websites, extract relevant data, and intelligently answer user queries based on the collected information
Responsibilities
- Implement robust pipelines using SerpAPI and FireCrawl to crawl and collect data from diverse websites and online sources
- Leverage Qdrant as a vector database for storing embeddings, ensuring high-performance retrieval based on user queries
- Utilize LangGraph and LangChain to build an intelligent agent system, orchestrating dynamic reasoning and context-aware responses. Mornitoring agent performance with LangSmith
- Employ Redis for maintaining conversation context, user query history, and session data, facilitating personalized and efficient AI interactions
- Develop and deploy a scalable FastAPI application
Technologies
- Python, FastAPI Framework
- LangChain, LangGraph and OpenAI
- Qdrant vector database and Redis
DOCUMENT AND CARD EXTRACTOR – JAPAN
AI – Backend Developer, May 2024 – September 2024
Project description
- Working on a team of 7 members including 1 Team Leader, 4 Developers and 2 QC.
- We were tasked to build main product ASAP AI backend and integrate AI services (Computer Vision, Natural Language Processing, LLMs) to enhance users’ experience
Responsibilities
- Analyse requirements, supported team to design software architecture, build and apply AI models
- Supported team to apply new technologies, implement plug-ins, services and handle issues occurred during development of the product.
- Research and fine tune AI models for specific tasks.
- Fixing bugs, supporting team members and maintaining.
- Ensure that deliveries are on-time and on-target.
- Clarify and Implement feature tasks.
- Participated in maintaining and developing new requests for customer
- Deploy the system for our customers.
- Documenting APIs and workflows
Technologies
- Python, Django Framework, RESTful for back-end
- OpenCV, Pillow, Scikit-learn, Scikit-image, Numpy, google-cloud-vision for Computer Vision tasks
- HuggingFace, OpenAI for NLP and LLMs tasks
- Visual Studio Code, Postman, Colab, Kaggle platforms
- PostgreSQL for storing data
- Nginx, Gunicorn, Uvicorn for deployment
- Jmeter for testing various cases of concurrent requests scenarios
- Swagger for documentation
E-LEARNING PLATFORM - VIETNAM
AI – Fullstack Developer, August 2023 – April 2024
Project description
- Working on a project of 5 member, including 1 Scrum Master, 2 QC, and 2 AI-Fullstack Developer
- The E-Learning Platform offers various educational courses. This website also provides AI features including Smart Chatbot, Image-Captioning, and Sentence Evaluation, improving users’ experience and helping children’s growth
Responsibilities
- Work with the lecturer to get the requirements and integrate the features.
- Receive requirements and develop the features.
- Research and fine-tune AI models for AI features
- Fix bug and maintain code
- Analyze requirements and seek for the optimal solutions
- Documenting the APIs and workflows
- Apply Gitlab CI/CD for auto integration
Technologies
- Python, Django Framework, RESTful for back-end
- JavaScript, ReactJS, Tailwind, Redux for front-end
- Langchain, Gemini for LLMs tasks
- PostgreSQL for storing data and Firebase for storing files
- Google Credentials for OAUTH2
- HuggingFace and Pytorch for Image-Captioning and Sentence Classification task
- Milvus – a vector database for storing vector data
- Visual Studio Code, Postman, Colab, Kaggle platforms
- Nginx, Gunicorn for deployment
- Redis as a broker for sending email, caching, being channel layer for Websocket
- Swagger for documentation
REAL-TIME COLLABORATION CHAT APPLICATION - VIETNAM
Backend Developer, October 2022 – July 2023
Project description
- Develop a real-time chat application that enables multiple users to exchange messages, share files, images, and participate in group channels.
- The application was deployed on AWS (ECS, API Gateway) to ensure scalability, performance, and reliability
- Implement FastAPI (async/await) combined with WebSockets to support real-time interactivity.
Responsibilities
- Analyze requirements and designed the chat system architecture using FastAPI’s asynchronous model and WebSockets.
- Integrate Redis as a message broker and caching layer to facilitate real-time communication between users.
- Secure APIs using token-based authentication (JWT) and implemented Amazon Cognito for user management and authorization.
- Optimize message sending/receiving performance to ensure low-latency chat operations.
- Integrate Celery for background task processing, such as push notifications, email sending
- Documented API and third-party integration guidelines using Swagger/OpenAPI.
Technologies
- FastAPI (async/await), Python
- Redis, Celery for background task processing and caching
- AWS: ECS, S3, API Gateway, Cognito
- Gitlab (CI/CD) for automated build, test, and deployment
- WebSockets for real-time communication
- Uvicorn to run FastAPI async applications
REAL-TIME EVENT TRACKING PLATFORM
Backend Developer, December 2021– September 2022
Project description
- Developed a real-time event collection and management platform that allows users to track, monitor, and analyze event data from various sources, including websites, mobile applications.
- Implemented FastAPI (async/await) as the backend framework to ensure high scalability, performance, and processing speed
- Stored event data in Amazon DynamoDB (NoSQL) to enhance query performance and efficiently handle high data volume and request loads.
- Deployed the application on AWS ECS, utilizing GitLab CI/CD for automated build, testing, and continuous deployment
Responsibilities
- Analyzed requirements and designed the real-time event tracking system architecture using FastAPI.
- Optimized the database structure in Amazon DynamoDB, designing tables and indexes to improve read/write performance.
- Secured APIs using Amazon Cognito to ensure proper authentication and authorization.
- Built real-time data ingestion functionality, processing and synchronizing events from multiple sources (webhooks, SDKs).
- Integrated Celery + Redis for background task processing, including event analysis, email/SMS notifications, and scheduled reporting.
- Automated build, testing (unit tests, integration tests), and deployment pipelines using GitLab CI/CD for staging and production environments.
- Monitored performance via AWS CloudWatch, optimizing ECS configurations and AWS resources to ensure stability and high availability.
- Created API documentation using Swagger/OpenAPI, along with installation and usage guidelines for clients and the QA team.
Technologies
- Amazon DynamoDB for NoSQL event data storage
- AWS: ECS, S3, Cognito, CloudWatch
- Celery + Redis for background task processing and caching
- GitLab CI/CD for automated build, test, and deployment pipelines
- Uvicorn for running FastAPI in an asynchronous environment
- Swagger/OpenAPI for API documentation and specifications
