Years of
Learning Code
Hello! I am An
Software Developer
Computer Science BSc student at ELTE with a research internship at Nokia Bell Labs and a peer-reviewed publication in Signals (MDPI). Experienced in ML pipelines (PyTorch), full-stack development (React, FastAPI), and deployment tooling (Docker).
0
0
Peer-Reviewed
Publication
Publication
0
Projects
Highlighted
Highlighted
0+
Technologies
Used
Used
My Strength
Computer Science student with a research internship at Nokia Bell Labs and a peer-reviewed publication. Hands-on experience building ML pipelines in PyTorch, full-stack applications with React and FastAPI, and containerized deployments with Docker. Strong analytical foundation from an honors BEng in Chemical Engineering.
STRENGTH
ML Research & Publication
During my research internship at Nokia Bell Labs (April - October 2025), I developed WiFi-CSI human activity recognition models in PyTorch, working with CNN architectures and exploring Vision Transformers (ViT). I built reproducible training and evaluation pipelines for the research team.
I co-authored a peer-reviewed journal paper published in Signals (MDPI, 2025) on dataset partitioning and data leakage in WiFi-CSI HAR, demonstrating its impact on reported performance. I also implemented post-training quantization with PyTorch FakeQuantize (1-8 bit) and benchmarked accuracy/F1, model size, and inference latency under different evaluation splits.
This research experience, combined with my honors Chemical Engineering degree from BME, gives me a strong analytical foundation for tackling complex problems with structured, data-driven approaches.
Key Skills
- PyTorch (CNN, Vision Transformers)
- Model Quantization (FakeQuantize, 1-8 bit)
- Data Analysis & Visualization (NumPy, Pandas, Matplotlib, Seaborn)
- Reproducible ML Pipelines
- Scientific Writing & Peer Review
- Signal Processing
STRENGTH
Full-Stack Development
At Nokia Bell Labs, I built an internal room monitoring dashboard using React, TypeScript, SCSS, and Recharts, integrating Express.js/JSON data sources with filtering and periodic refresh. I also deployed inference services via FastAPI.
For GlowBeautyBox, I worked on an online booking system using Wix, HTML, CSS, and JavaScript, and helped build an internal calendar and staff management system with Python Django, utilizing API calls to manage bookings and schedules.
I've also built websites from scratch (HTML/CSS/JS), a restaurant website, and this portfolio site, gaining experience across the full web development stack.
Key Skills
- React, TypeScript, SCSS
- Node.js, Express.js
- FastAPI, Django
- HTML/CSS, JavaScript
- REST API Design & Integration
- Responsive Web Design
STRENGTH
Software Engineering (C#, Java, Python)
C# .NET:
Developed a microservice-based file management system using C# .NET and DevOps principles. Implemented RESTful GET/POST APIs for real-time file content serving with change detection and data persistence. Built a relay service for scheduled data retrieval and conditional updates with transactional logging.
Java:
Created board game simulations emphasizing OOP principles (inheritance, polymorphism, design patterns) with comprehensive JUnit 5 testing. Applied Singleton and Visitor patterns for complex simulations.
Python:
Built ML pipelines in PyTorch, preprocessing tooling with NumPy/Pandas, FastAPI backend servers, and desktop UIs with Tkinter. Experienced with scikit-learn for data analysis.
Database:
Designed relational schemas with normalization (1NF to BCNF), wrote Oracle SQL queries for data retrieval and aggregation, and PL/SQL procedures and triggers for automated data handling.
Key Skills
- C#, .NET, RESTful API Design
- Java, JUnit 5, Design Patterns
- Python, PyTorch, FastAPI
- Oracle SQL, PL/SQL
- OOP, Algorithms & Data Structures
- Unit Testing
STRENGTH
DevOps & Deployment
At Nokia Bell Labs, I containerized ML tooling and applications with Docker, targeting both GPU machines and Raspberry Pi deployments. I contributed to debugging platform and hardware compatibility issues across different deployment targets.
I use Git/GitHub for version control across all projects, with experience in CI/CD pipelines using GitHub Actions. I'm comfortable working from the command line with Bash and shell scripting for automation and development workflows.
Key Skills
- Docker (Containerization, Multi-platform)
- Git/GitHub (Version Control, Collaboration)
- Automation Development
- Bash/Shell Scripting
- Raspberry Pi Deployment
- Command Line (CLI) Proficiency
My Recent Works
Here I include relevant school and individual projects that I have done so far.
ML / Research
WiFi-CSI Human Activity Recognition
PyTorch CNN/ViT models, post-training quantization, peer-reviewed publication at Nokia Bell Labs.
Web / ML
Room Monitoring Dashboard
React, TypeScript, SCSS, Recharts dashboard with Express.js integration at Nokia Bell Labs.
Backend
File Management System (C# .NET)
Microservice-based RESTful API for automated file processing and data synchronization.
Python / Web
GlowBeautyBox Booking System
Online booking and staff management system using Django, Wix, and API integration.
Web / PHP
Car Rental System (PHP)
Dynamic car rental system with user/admin roles, booking, and price calculation.
Database
Database Management - SQL & PL/SQL
Relational schema design, normalization, query optimization, and PL/SQL procedures.
VietFood Recipe Website
This is an University Project, required by "Web Development" Subject in the 2nd Semester at ELTE University. This project was build individually.
Category
Web Design
Purpose
University Assignment
Start Date
May, 2024
Project Description
This project utilized HTML/CSS and a bit of JavaScript to build a 4-page website with the function of storing recipes.
Elaboration
Being a foreign Vietnamese student in Hungary, I crave a lot of food from my hometown. Therefore this website was made to tribute Vietnamese food. This website is also hosted locally in the University system, which could be accessed through the University Wifi at ELTE University. This website does include JavaScript mainly to help with Searching Function, slideshow of Images and the Form Submission.
RELATED
Github was mainly used for version control.
Restaurant Website
A website with 2 languages I made for a Vietnamese Restaurant located in Budapest, Hungary. I made with the help of Wix - a web development services.
Category
Web Design
Purpose
Personal Project
Start Date
June, 2024
Project Description
This project utilized HTML/CSS and a bit of JavaScript to build a 4-page website with the function of storing recipes.
Features
Fully functioned website, including Vietnamese and Hungarian version. Menu, About, Gallery, Downloading Menu and Reservation Pages included.
More
Under development - further features intended:
Fast API server
This is an University Project that belongs to subject Python. In this project, I worked on completing and refining a FastAPI backend server that was partially developed in a previous assignment. The primary goal was to finalize the implementation of the remaining components and ensure the backend application is fully functional and operational.
Category
Python
Purpose
University Project
Start Date
March, 2024
Project Description
Main Tasks Finished
Cloned the project repository from GitHub and set it up locally.
Implement API Endpoints:
- get_all_events: Retrieve all events from event.json.
- get_events_by_filter: Filter events by date, organizer, status, and event type.
- get_event_by_id: Retrieve an event by its ID.
- create_event: Add a new event to event.json.
- update_event: Update an existing event by its ID.
- delete_event: Delete an event by its ID.
- get_joiners_multiple_meetings: Retrieve joiners who attended at least two meetings.
Testing and Debugging.
Features
- User Authentication and Authorization: Implemented JWT-based authentication to secure the API endpoints.
- CRUD Operations: Developed Create, Read, Update, and Delete functionalities for the primary resources.
- Error Handling: Added custom error handling to provide meaningful error messages and HTTP status codes.
- Data Validation: Used Pydantic models for request data validation and response formatting.
- API Documentation: Leveraged FastAPI’s built-in automatic documentation generation (Swagger UI).
Gas Layer Simulation Project
This project, part of my university coursework in subject Object-Oriented Programming, involves simulating the behavior of atmospheric gas layers under various conditions. The goal was to create a system where different gas layers (ozone, oxygen, carbon dioxide) interact with atmospheric variables (thunderstorm, sunshine, other effects), transforming and ascending accordingly. The project emphasizes the application of OOP principles, design patterns, and comprehensive unit testing to ensure a robust and maintainable codebase.
Category
OOP
Purpose
University Project
Start Date
March, 2024
Project Description
Main Tasks Finished
To complete the assignment, I developed a simulation that:
- Reads data from an input file specifying gas layers and atmospheric conditions.
- Uses a collection of gas layers, each represented by a class that inherits from a common base class and overrides virtual methods.
- Simulates the interaction of atmospheric variables with the gas layers, applying transformations based on specific rules.
- Utilizes the Singleton and Visitor design patterns to manage the simulation's state and operations.
- Ensures no gas layer has a thickness less than 0.5 km unless it ascends and merges with an identical layer above it.
- Continues the simulation until one type of gas layer perishes completely from the atmosphere.
- Prints the state of the gas layers after each simulation round.
Features
- Inheritance and Polymorphism: Implemented a hierarchy of gas layer classes inheriting from a base class.
- Collection Management: Stored gas layer objects in a collection and traversed it to apply transformations.
- Singleton Pattern: Ensured a single instance of the atmospheric conditions manager.
- Visitor Pattern: Applied transformations to gas layers based on atmospheric conditions.
- Exception Handling: Incorporated robust exception handling to manage invalid operations and ensure program stability.
- File I/O: Read input data from a file and printed the state of gas layers after each simulation round.
- Unit Testing: Developed comprehensive unit tests to validate the functionality of each component.
- Modular Design: Divided the program into modules for better organization and maintainability.
ChessBoard School Assignment
This is a university project for the subject Java. In this project, I developed and tested a complex board game simulation. The primary focus was on implementing core OOP principles, adhering to Java coding standards, and utilizing JUnit 5 to ensure code quality and functionality.
Category
OOP
Purpose
University Project
Start Date
May, 2024
Project Description
Main Features
- Board Setup: Developed a
WalkingBoardclass to represent a game board with customizable sizes and values. - Movement & Operations: Added methods to get the board’s state, move the figure, and ensure moves stay within bounds.
- Extended Functionality: Enhanced the board to support multiple players, including a special player type with unique behavior.
- Testing: Used JUnit 5 to test board functions and player interactions thoroughly.
Highlights
- Flexible Initialization: Supported various board sizes and values, including non-standard row lengths.
- Special Player Features: Implemented a unique player with advanced movement and scoring rules.
- Thorough Testing: Ensured code reliability with a complete set of tests for all key functions.
- Clean Code: Followed Java best practices for clear and maintainable code.
Calendar System Management
Glow Beauty Box is a beauty franchise in Budapest, specializing in pedicures, manicures, and related services. The owner sought an efficient booking system and calendar management tool to handle the extensive details of multiple stores and staff members. Under the guidance of a senior developer, I contributed to the development of a booking system and implemented a calendar management system for staff and store schedules.
Category
Python
Purpose
Other Project
Start Date
May, 2024
Project Description
The booking system was implemented using the built-in functions of Wix. To address the complexity of managing multiple staff members and store locations, a linked staff and store management system was developed. This system, which integrates seamlessly with the existing Wix platform for calendar setup and scheduling, was built using the Django framework, along with HTML, CSS, and JavaScript.
Booking System
- Designed and launched an online booking system for Glow Beauty Box using Wix, improving client accessibility.
- Integrated automated notifications reducing no-shows
- Customized UI for a seamless user experience aligned with the brand.
- Enhanced mobile responsiveness and integrated secure payment options.
- Provided ongoing maintenance to ensure system reliability and security.
Calendar Manangement
- Custom Staff Management System: Developed using Django, HTML, CSS, and JavaScript, improving scheduling and personnel management.
- Dynamic Calendars: Created publicly accessible calendars, ensuring real-time visibility of work schedules.
- Streamlined Scheduling: Reduced manual errors and improved operational efficiency.
- Role-Based Access Controls: Implemented to secure and manage sensitive staff information.
- Responsive Interface Design: Ensured easy schedule management across all devices.
- Seamless Updates: Enabled accuracy and timeliness in calendars and staff information.
WiFi-CSI Human Activity Recognition
Research internship at Nokia Bell Labs developing WiFi-CSI human activity recognition models in PyTorch. Co-authored a peer-reviewed publication in Signals (MDPI) on evaluation integrity and data leakage in HAR systems.
Category
ML / Research
Purpose
Research Internship
Duration
April - October 2025
Project Description
Developed WiFi-CSI human activity recognition models in PyTorch (CNN; explored ViT) and built reproducible training/evaluation pipelines. Implemented post-training quantization with PyTorch FakeQuantize (1-8 bit) and benchmarked accuracy/F1, model size, and inference latency under different evaluation splits.
Key Contributions
- Co-authored peer-reviewed paper: Varga, D.; Cao, A.Q. "Why Partitioning Matters: Revealing Overestimated Performance in WiFi-CSI-Based Human Action Recognition." Signals, 2025, 6, 59.
- Built preprocessing and analysis tooling (NumPy, Pandas) and generated research visuals (Matplotlib/Seaborn, confusion matrices, ablations).
- Deployed inference via FastAPI; built a Python desktop UI (Tkinter/ttkbootstrap) for RSSI logging and signal visualization.
- Containerized ML tooling with Docker for GPU machines and Raspberry Pi targets.
Technologies
- PyTorch, CNN, Vision Transformers (ViT)
- FakeQuantize (1-8 bit quantization)
- NumPy, Pandas, Matplotlib, Seaborn, scikit-learn
- FastAPI, Docker, Tkinter/ttkbootstrap
Room Monitoring Dashboard
Internal room monitoring dashboard built at Nokia Bell Labs, integrating Express.js/JSON data sources with filtering and periodic refresh.
Category
Web / Full-Stack
Purpose
Nokia Bell Labs Internal Tool
Duration
2025
Features
- Built with React, TypeScript, and SCSS
- Data visualization using Recharts
- Integration with Express.js/JSON data sources
- Filtering and periodic data refresh
API-based File Management System (C# .NET)
Microservice-based system using C# .NET and DevOps principles for automated file processing and data synchronization.
Category
Backend / OOP
Purpose
University Project
Start Date
February 2025
Key Features
- RESTful GET API: Serves real-time file content with change detection.
- POST API: Handles file updates and persists data.
- Relay Service: Scheduled data retrieval and conditional updates with transactional logging and continuous operation.
Car Rental System (PHP)
Dynamic car rental system with user and admin roles, car booking with date selection, price calculation, and booking management.
Category
Web / PHP
Purpose
University Project
Start Date
January 2025
Features
- User and admin role management
- Car booking with date selection and price calculation
- Booking management system
- PHP with JSON storage and custom CSS for UI
Interactive Railway Board Game
JavaScript board game with timer, leaderboard, difficulty levels, responsive UI, and tile rotations for an interactive gameplay experience.
Category
Web / JavaScript
Purpose
University Project
Start Date
November 2024
Features
- Timer and leaderboard system
- Multiple difficulty levels
- Responsive UI design
- Tile rotation mechanics
- Built with JavaScript, HTML, and CSS
Database Management - SQL & PL/SQL
Designed relational database schemas and implemented SQL queries, PL/SQL procedures, and triggers for automated data handling and reporting.
Category
Database / Backend
Purpose
University Project
Start Date
2024
Key Topics
- Relational schema design with normalization (1NF to BCNF)
- SQL queries for data retrieval, aggregation, and reporting
- PL/SQL procedures and triggers
- Query optimization techniques
- Indexing (B+ trees, linear hashing)
My Experience
My Education
My Skills
What I have learnt so far!
75%
Python
60%
PyTorch
50%
React / TypeScript
50%
C# / .NET
50%
Docker
50%
JavaScript
45%
Oracle SQL
45%
Java
55%
HTML/CSS
55%
Git
35%
Node.js
20%
C
Let’s work together!
-
Phone
+36 70 674 9906 -
Email
ancaoquynh@gmail.com -
Based in
Budapest, Hungary