Skip to content
View Prhm93's full-sized avatar
4 followers · 14 following

Block or report Prhm93

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Prhm93/README.md

Hi there 👋, I'm Parham

AI Engineer Computer Vision Deep Learning Image Segmentation Object Detection Agentic AI Data Science GIS & Spatial AI

👤 About Me

I'm an AI Engineer and researcher specialising in computer vision, deep learning, and applied ML systems. My background spans civil engineering and computer science — I build end-to-end AI pipelines that solve real-world problems, from model architecture through to deployed applications.

I work across the full stack: designing neural network architectures, training and evaluating models, and shipping production-ready systems.

🛠 Skills & Stack

AI / ML
PyTorch TensorFlow Hugging Face Transformers Scikit-learn OpenCV

Computer Vision & Detection
Semantic Segmentation Object Detection (YOLO) Sliced Inference (SAHI) Albumentations

LLMs & Agentic AI
Prompt Engineering RAG Tool Use / Function Calling Hugging Face

Backend & Data
FastAPI PostgreSQL PostGIS MySQL

Frontend & GIS
React Leaflet Geospatial Visualisation

DevOps & Environment
Docker Git Linux Google Colab

🚀 Projects

FloodWatch

Full-stack AI application for real-time flood scene analysis.
Three-stage pipeline: semantic segmentation → human detection → emergency classification (EMERGENCY / MONITOR / SAFE)
FastAPI PostgreSQL/PostGIS React Leaflet Docker PyTorch

TMSA-Net

Transformer-based Multi-Scale Attention U-Net for flood-water segmentation.
EfficientNet-B4/ResNet34 encoder · Transformer bottleneck · Multi-scale attention gates · ~92% IoU
PyTorch Semantic Segmentation Transformers

DKAG U-Net

Dynamic Kernel Attention Gate — replaces fixed 1×1 attention projections with adaptive depthwise-separable convolutions. +3.41 IoU over standard U-Net baseline. Presented at IPGRC 2026.
[GitHub →]

📬 Get in Touch

LinkedIn GitHub Email

Popular repositories Loading

  1. hands-on-python-3084712 hands-on-python-3084712 Public

    Forked from LinkedInLearning/hands-on-python-3084712

    This is a repository for the LinkedIn Learning course Hands-On Python

    HTML 2

  2. python-for-data-science-and-machine-learning-essential-training-part-1-3006708 python-for-data-science-and-machine-learning-essential-training-part-1-3006708 Public

    Forked from LinkedInLearning/python-for-data-science-and-machine-learning-essential-training-part-1-3006708

    This is a repository for the LinkedIn Learning course Python for Data Science and Machine Learning Essential Training

    Jupyter Notebook 2

  3. Visual-Information-Analysis Visual-Information-Analysis Public

    Jupyter Notebook 2

  4. rag-insurance-claims-analysis rag-insurance-claims-analysis Public

    RAG pipeline for querying motor insurance documents using FAISS and Google Gemini

    Jupyter Notebook 2

  5. dkag-unet-flood dkag-unet-flood Public

    Dynamic Kernel Attention Gate U-Net for flood image segmentation (IPGRC 2026)

    Python 2

  6. Python-Programming Python-Programming Public

    Forked from pytopia/Python-Programming

    Python Programming With Ali Hejazi (Pytopia)

    Jupyter Notebook 2