I'm an AI-focused developer interested in building intelligent systems and understanding the technical foundations behind them.
My current focus is on AI Agents, Computer Vision, Deep Learning, and Machine Learning.
An AI-powered commerce system for sellers, currently starting with Telegram.
The AI Agent can interact with customers, access store data, answer product questions, and assist with order processing through tool calling.
Tech:
Python · FastAPI · LLM API · AI Agent · Tool Calling · Telegram Bot · Telegram Mini App · SQLite
The backend is gradually moving toward a Feature-based Modular Monolith / Vertical Slice Architecture.
A computer vision project focused on building a practical real-time pipeline for object detection, multi-object tracking, trajectory analysis, and object counting.
The project currently combines YOLO + ByteTrack and explores the underlying concepts behind modern detection and tracking systems, including:
- Bounding boxes and confidence scores
- IoU
- Non-Maximum Suppression (NMS)
- Class-aware NMS
- Video-based detection
- Multi-object tracking
- Track IDs
- Detection and processing performance
The project is being developed incrementally toward trajectory analysis, counting, evaluation, and robustness analysis.
Tech:
Python · PyTorch · YOLO · OpenCV · ByteTrack · Jupyter · Google Colab
I'm also building and experimenting with computer vision projects using PyTorch.
My focus includes:
- Image classification
- CNNs
- Dataset analysis
- Model evaluation
- Error analysis
- Object detection
I'm particularly interested in understanding why models make mistakes and how their performance can be improved.
- Artificial Intelligence
- Machine Learning
- Deep Learning
- Computer Vision
- AI Agents
- LLM-based Systems
- Mathematics for Machine Learning
- Algorithms & Optimization
- Robotics and Intelligent Systems
YOLO + ByteTrack pipeline for real-time object detection and multi-object tracking, with a focus on understanding detection, tracking, and evaluation.
AI-powered commerce assistant combining LLM APIs, AI Agents, tool calling, Telegram, and a FastAPI backend.
PyTorch-based experiments in image classification, CNNs, dataset analysis, and model error analysis.
A Persian question-answering project built around a vehicle manual, with source-aware answers and structured retrieval.
I'm continuously working on improving my understanding of:
- Deep Learning fundamentals
- Computer Vision
- Neural network architectures
- Object Detection & Tracking
- Model optimization
- Machine Learning mathematics
- AI Agent architectures
- Intelligent systems
Build. Experiment. Understand. Improve.