Develop technology to make life better

Here is my technology repo. Anyone can use my work. Feel free to use it to help others.

---

I. Projects

Drone Vision

Advanced drone vision system with trajectory planning, object detection, and visual SLAM.

• Drone control using ROS2
• Customized YOLOv5
• Drone-Visual SLAM in farming
• Point Cloud Processing

developed at hexafarms

Computer Vision ROS2 SLAM

3D Reconstruction

Advanced 3D reconstruction system for creating detailed digital models.

developed at hexafarms

Computer Vision 3D Modeling SfM

Data Science

Advanced data science solutions for complex problems.

• Image Segmentation
• Data Processing
• Visualization

developed at hexafarms

Data Science ML Visualization

Image Processing

• Instant Segmentation
• Image-tabular data combined Classification
• Time-series analysis for growth

developed at hexafarms

Computer Vision Classification Segmentation

LLM Powered Web Chatbot

• Natural Language Processing, Word Embedding
• Design Concurrency Functionality
• Prompt Engineering

developed at hexafarms

NLP LLM AI

AI Cover Letter Generator

• Multi-Agent Workflow for comprehensive letter generation
• RAG (Retrieval-Augmented Generation) for company research
• Intelligent Skill Extraction & Matching

Streamlit-based web application that uses LLMs to generate personalized cover letters by analyzing job descriptions and researching companies in real-time.

LangChain RAG Multi-Agent

MLOps & HyperParameter Tuning

• Fully Open source-based full MLOps (Version, feature control)
• Hyper parameter tuning with experiment tracking

MLOps Model Tuning Experiment Tracking

Time-series Forecast

• Develop domain knowledge-supported custom ML MultiConstrainedLinearRegression
• Feature Engineering/Imputation (Transformer/GBT-based, statistic, plant science)
• XAI (Explainable AI) to understand why model performs better or worse

developed at hexafarms

Time Series Forecasting XAI

Stereo Vision

• Build hardware of stereo vision from scratch
• Depth computation
• Reading

developed at hexafarms

Computer Vision 3D Reconstruction Depth Estimation

---

II. More Source Codes

AI Cover Letter Generator

Multi-agent workflow with RAG capabilities for personalized cover letter generation using LLMs.

View on GitHub

MultiConstrained LinearRegression

Advanced regression algorithms with multiple constraints for scientific applications.

View on GitHub

Computer Vision Ops

Operations and utilities for computer vision applications.

View on GitHub

CharucoStereoCalibrator

Tool for calibrating stereo camera systems using Charuco boards.

View on GitHub

Berlin Reservation Bot

Automated system for booking appointments with Berlin authorities.

View on GitHub

More Projects

• Compute pixel's actual dimension
• YOLO data generator
• Image classification & mass estimation
• Image cropping tool
• Image Segmentation using Graph-cut
• MLflow-HyperOpt wrapper
• Science into Code
• Datumaro GUI app
• Classifier Wrapper Platform
• Custom Chain of Thought using Gemini

---

III. Publications

Master Thesis

Deep Learning-Based Semiautomatic Generation of HD maps from Aerial Imagery

Deep Learning HD Maps Aerial Imagery

Deep Aerial Mapper

Research paper on advanced aerial mapping techniques using deep learning.

View Paper

Research Computer Vision arXiv

---

IV. Learning

1. Camera Parameter
2. Point Cloud Library
3. 3D Reconstruction
4. Color and Illumination
5. Feature extraction
6. Plant nutrient deficiency
7. Public Leaf dataset
8. Plant biology
9. ROS2, Udemy
10. Advanced C++, Coursera
11. Data Structure
12. Airflow, Udemy
13. Open3d
14. Elixir, Phoenix

I learn hard and apply the knowledge assiduously. But I still relish chill, too.