ABOUT ME

AI & Data Enthusiast

Hi, I’m Hnin Ei San.

I recently graduated in Robotics & AI Engineering from KMITL, and along the way I gained experience through internships at Ford Motor Company and Infineon Technologies.

I build end-to-end AI products that actually feel good to use where both the user experience and the code behind it are thoughtfully designed. I enjoy working across data analytics, machine learning, computer vision, deep learning, LLMs, and RAG pipelines basically, whatever the problem needs.

Currently based in Bangkok, Thailand and open to full-time roles and international opportunities.

HNIN EI SAN
  • FOCUSAI, Data Science, Computer Vision
  • EDUCATIONB.E. Robotics & AI @ KMITL
  • LOCATIONBangkok, TH
  • LANGUAGESEnglish, Burmese

FEATURED WORK

Projects & Builds

Energy Consumption Analysis 1 Energy Consumption Analysis 2
BIG DATA
Energy Consumption Analysis for Drone Delivery
Analyzed and predicted energy consumption of drones during package delivery missions using big data processing pipelines and statistical modeling.
Python Big Data Pandas Matplotlib
↗
AI Debugging Assistant 1 AI Debugging Assistant 2
AI / DATA SCIENCE
AI Debugging Assistant
Automatically detects, explains, and fixes bugs in source code using a LangGraph multi-agent pipeline with LLM reasoning.
Python LangGraph LLMs Agentic AI
↗
Transformable Robotic System 1 Transformable Robotic System 2
ROBOTICS
Transformable Multi-Modal Robotic System
Transformable hybrid UAV-UGV platform capable of aerial flight, ground locomotion, and mechanical morphing for visual inspection tasks.
ROS2 Python C++ PX4 STM32 Deep Learning Computer Vision
↗
Automated pH Estimation 1 Automated pH Estimation 2
COMPUTER VISION
Automated pH Estimation
Automated pH level estimation of water samples by analyzing color changes on test strips
Python OpenCV KNN K-means clustering
↗
Vision Transformer 1 Vision Transformer 2
MACHINE LEARNING
Vision Transformer & MLP-Mixer
Implemented standard Transformer to Supervised Pre-Training for image classification and Self-Supervised Pre-Training for segmentation tasks.
PyTorch ViT MLP-Mixer Self-Supervised
↗
LLM & RAG 1 LLM & RAG 2
AI
AI-Enabled Framework for Strengthening University–Industry Collaboration
The study proposes an AI-enabled UIC system with two stages: a pre-collaboration tool that matches suitable partners and evaluates their credibility, and a post-collaboration system that provides structured guidance for implementation.
Recommendation systems Innovation ecosystems Credibility assessment
↗
Height Measurement 1 Height Measurement 2
COMPUTER VISION
Height Measurement from a Single Camera View
Developed a computer vision system for measuring height from the single camera perspective using monocular depth estimation models.
OpenCV NumPy Pose Estimation Object Detection Flask
↗

TECHNICAL STACK

Skills & Tools

Programming

Python
C
SQL
MATLAB

AI & ML

Machine Learning
Deep Learning
Computer Vision
LLMs
RAG
NLP
Agentic AI
Crew AI
GenAI

Data Analysis

Data Cleaning
EDA
Data Transformation
Statistical Analysis

Visualization

Matplotlib
Seaborn
Power BI

Tools & Infrastructure

MATLAB
Excel
Git
Linux (Ubuntu)
Docker
Stream-lit
SharePoint
Power Automate

ACADEMIC BACKGROUND

Education

KMITL Logo

Bachelor of Engineering in Robotics & AI

King Mongkut's Institute of Technology Ladkrabang (KMITL)

2022 — 2026

↗

RECOGNITION

Honors & Awards

Freshman Scholarship Recipient(2022-2026)

King Mongkut’s Institute of Technology Ladkrabang

AI for ASIA Fellowship 2025

Siklab (Skills for Asia)

KAIST EE Camp 2025

Korea Advanced Institute of Science and Technology, Electrical Engineering

National Finalist, Myanmar 2024

ASEAN Data Science Explorer

On Campus Winner 2024

Hult Prize

Tourism Case Winner 2022

International Summer School and Hackathon on Technology and Sustainability, Thailand

REACH OUT

Let's Connect

OPEN TO OPPORTUNITIES & COLLABORATIONS

hninei.em@gmail.com
BIG DATA CASE STUDY

Energy Consumption Analysis for Drone Delivery

This project is designed to predict the energy requirements of autonomous drones during package delivery lifecycles. By integrating high-fidelity telemetry data with physical and statistical modeling, the system achieves an understanding of flight dynamics and battery performance.

Technical Overview

  • Automatically detect takeoff, cruise, and landing phases using altitude-time profiles and Gaussian filtering.
  • Compute induced power and other energy-related metrics based on flight dynamics and environmental conditions.
  • Train a linear regression model that correlates physical power requirements with measured battery usage.
  • Estimate energy consumption for test flights and evaluate accuracy using Average Relative Error (ARE).

Analytics & Results

  • Evaluated CO₂ reduction potential, demonstrating drones as a sustainable alternative for last-mile delivery.
  • Achieved high precision regression results, validated against real-world test flight data using ARE metrics.
  • Identified key drivers of battery discharge, providing insights into how payload affects energy usage.
  • Developed a pipeline for processing telemetry data, enabling better path planning for autonomous missions.
Methodology Diagram
Data Variables Table
Results Chart
GHG Results Chart
King Mongkut's Institute of Technology Ladkrabang

Robotics & AI Engineering

Highlights

  • KMITL Freshman Scholarship Recipient
  • Teacher Assistant @ Physics Lab I
  • Teacher Assistant @ Physics Lab II
  • Teacher Assistant @ Introduction to AI Course
  • GPA : 3.64/4.00

Coursework

  • Machine Learning in Practice
  • Computer Vision
  • Introduction to Data Science
  • Big Data
  • Artificial Intelligence Technology
  • Mathematics of Data Science
  • Probability and Statistics
  • Differential Equations & Matrix Algebra
  • Introduction to Algorithms
  • Information Systems & Databases
  • Mobile Application Development
  • Microprocessor & Interface
  • Robotics Laboratory
AGENTIC AI CASE STUDY

AI Debugging Assistant

Modern software debugging is often time-consuming. This assistant leverages a multi-agent orchestration framework using LangGraph to automate the detection, analysis, and fixing of bugs within Python applications, simulating a senior developer's reasoning process.

Technical Overview

  • Orchestrates specialized agents (Scanner, Explainer, Fixer) utilizing LangGraph for complex iterative reasoning.
  • Employs advanced LLMs to perform deep semantic analysis of codebases to identify subtle logical errors.
  • Generates and validates precise code fixes, ensuring high reliability through integrated verification.
  • Uses feedback loops to continuously improve fix suggestions until the detected bug is resolved.

Analytics & Results

  • Significantly reduced debugging time for complex Python applications by providing clear explanations and fixes.
  • Successfully processed multi-file repositories with interdependent modules, demonstrating the robustness of the approach.
  • High success rate in identifying and fixing edge cases that traditional static analysis tools often miss.
Senior Capstone Project

Transformable Multi-Modal Robotic System

Development of a hybrid UAV-UGV platform capable of seamless transitions between aerial and terrestrial locomotion for enhanced operational versatility in complex environments.

poster
Poster
Transformation Mechanism
Physical transformation mechanism engaged during UAV-to-UGV transition.
Embedded Controller
Custom embedded control board for multi-modal operation.
COMPUTER VISION CASE STUDY

Automated pH Estimation

An automated system for pH level estimation using computer vision and machine learning, designed to provide accurate and real-time analysis of chemical indicator strips.

Technical Overview

  • Developed a CV-based pipeline that automatically segments pH indicator strips and extracts precise color values from test regions.
  • Implemented a robust normalization algorithm that uses a reference color chart to account for varied lighting conditions.
  • Trained regression models to map extracted RGB/HSV color vectors to precise pH values, achieving high numerical accuracy.
  • Optimized the image processing workflow to enable real-time pH estimation on embedded devices and mobile platforms.

Analytics & Results

  • Achieved a mean absolute error (MAE) of less than 0.1 pH units, consistently outperforming human subjects in controlled testing.
  • Successfully validated the system's performance across varied lighting environments through advanced white balancing.
  • Demonstrated the potential for low-cost, automated chemical analysis in agricultural monitoring applications.
Color Segmentation
Automatic segmentation of indicator regions from laboratory image data.
pH Regression Results
Regression model performance across the full pH spectrum.
COMPUTER VISION CASE STUDY

Height Measurement from a Single Camera View

A computer vision system designed to accurately estimate the height of objects or individuals from a single calibrated camera view using geometric perspective and reference objects.

Technical Overview

  • Implemented a geometric reasoning engine that utilizes vanishing points and horizon line detection to project real-world heights.
  • Developed an automated calibration routine using objects of known height to establish pixel-to-metric conversion factors.
  • Integrated YOLO-based models to automatically detect and segment individuals or objects for measurement within the frame.
  • Developed noise-reduction filters to account for lens distortion and camera tilt, ensuring stability across viewing angles.

Analytics & Results

  • Achieved a deviation of less than 3% compared to physical measurements at distances up to 10 meters.
  • Optimized for low-latency performance, enabling real-time height estimation for surveillance applications.
  • Demonstrated a streamlined setup process requiring only a single reference object for calibration.
Geometric Construction
Visualizing the geometric perspective lines and horizon detection for height inference.