Embedded AI & IoT Systems Engineer

PHAN THI QUYNH TRAM

********************@*****.***

078*******

Ho Chi Minh City

https://github.com/MiaSa8052

ABOUT ME

PROJECTS

Final-year Electronic Physics student with strong skills in programming, embedded systems, and AI/computer vision. Experienced in research and teamwork, co-developing a deep learning framework for rice leaf disease detection accepted at ICEBA 2026. Motivated, adaptable, and eager to contribute to innovative projects in embedded systems and AIoT.

EDUCATION

University of Science – Ho Chi Minh City

(2022 - now)

Major: Electronic Physics

GPA: 3.15 / 4.0

B1 VSTEP

Basic IT Application skills certificate

Rice Leaf Disease Detection System Research Project Role: AI Developer & Full-stack Developer (Collaborative R&D Team)

TECHNICAL SKILLS

1.Programming Languages: C/C++,

Python, Assembly, Verilog HDL.

2.Embedded Systems & IoT: ESP32-AI,

ESP8266, Arduino, Jetson Nano,

FPGA, ATmega128.

5.AI & Computer Vision: YOLO,

SAHI, Image processing, Machine

learning.

4.Control & Processing: Control logic,

Sensor signal processing, Actuator

control.

6.Other skills: Teamwork & Collaboration,

Technical Documentation, Problem-Solving,

Scientific Research, Active Learning.

3.Automation & Simulation: Proteus,

Wokwi, Microchip Studio, Arduino

IDE, Quartus, VS code, Docker, Git.

Technologies used: Python, PyTorch, YOLOv8, SAHI,

WBF, HTML/CSS/JS, JSON.

Description: Collaborated in a research team to develop and compare independent deep learning pipelines, achieving an enhanced framework for rice leaf disease diagnosis submitted to the ICEBA 2026 conference.

Key Responsibilities & Achievements:

Data Pipeline: Managed and annotated a 15,000-image dataset (cleaning, manual/semi-automated labeling). Model Training: Fine-tuned YOLOv8 on Google

Colab; integrated SAHI to enhance small lesion

detection.

Custom Post-processing: Replaced NMS with a

custom pipeline using Weighted Boxes Fusion (WBF)

to resolve box overlapping.

Deployment: Built a lightweight Web App

(HTML/CSS/JS) to serve real-time predictions and

solutions via JSON.

Result: Successfully optimized the system to achieve a peak F1-score of 0.837, significantly enhancing the sensitivity for early-stage micro-lesion detection.

Contact this candidate