Zafir Abdullah

Building production-ready intelligent systems that bridge the gap between research and real-world impact.

5+ Projects Shipped
3+ AI Domains
Open to Opportunities

Who I Am

I'm an AI Developer based in Karachi, Pakistan, passionate about designing intelligent systems that solve real problems from real-time surveillance to generative AI pipelines.

My work spans the full ML lifecycle: data cleaning, model development, evaluation, and deployment. I enjoy working at the intersection of research and engineering taking ideas from paper to production.

Currently pursuing studies at Virtual University of Pakistan while building hands on experience through projects and exploring MS-AI as a future goal.

Current Focus

  • ⚡ Real-Time AI Surveillance & Vision Systems
  • 🤖 Generative AI & Agentic AI Pipelines
  • ☁️ Scalable MLOps & Cloud Deployment

Tech Stack

AI / ML Frameworks

Python PyTorch TensorFlow Keras scikit-learn OpenCV YOLOv8 Hugging Face

Data & ML Engineering

Pandas NumPy XGBoost LightGBM Optuna MLflow NLTK spaCy

Backend & APIs

FastAPI Flask Django Streamlit SQLite PostgreSQL MongoDB

Cloud & DevOps

AWS EC2 Azure GCP Docker Linux Git GitHub Actions

Frontend

JavaScript TypeScript HTML / CSS Angular

Core Language Proficiency

Python 75%
JavaScript 50%
TypeScript 50%
HTML / CSS 50%

Featured Work

NLP SaaS

AI Resume Screening SaaS

NLP-based resume screening and candidate matching platform. Automates talent discovery by intelligently parsing resumes and ranking candidates against job descriptions.

NLP Python spaCy FastAPI
Deep Learning CV

Real-Time Person ReID

Multi-camera real-time person re-identification system. Tracks individuals across different camera feeds using deep feature extraction and similarity matching.

PyTorch OpenCV Deep Learning YOLO
Transfer Learning

Stress Detection · Facial AI

Facial stress classification using MobileNetV2 transfer learning. Deployed as an interactive Streamlit app with real-time webcam inference and confidence scoring.

MobileNetV2 TensorFlow Streamlit OpenCV
In Progress NLP

Reddit Sentiment Analysis

End-to-end ML pipeline with ensemble stacking (XGBoost + LightGBM + Random Forest), Optuna hyperparameter tuning, imbalanced data handling, and MLflow experiment tracking on AWS EC2.

XGBoost LightGBM Optuna MLflow AWS EC2
Segmentation

U-Net Image Segmentation

Semantic image segmentation using U-Net architecture. Trained on custom datasets with data augmentation for precise pixel-level classification.

U-Net PyTorch OpenCV

Education

BS-Business and Information Technology

Virtual University of Pakistan

Current

Pursuing undergraduate studies with coursework spanning business management, operations, quality management, and business law complemented by self-directed IT skill building.

CS201 CS610 CS614

MS-Computer ScienceGoal

Future

Targeting the MSCS program with specialization in Machine Learning and Deep Learning. Core curriculum includes Advanced AI, Mathematical Foundations of AI, Advanced ML, and Research Methodology.

Advanced AI Deep Learning Research

Let's Build Together

Open to internships, entry-level roles, freelance collaborations, and research opportunities.

Location

Karachi, Pakistan · Remote Friendly

This form opens your email client. You can also reach out directly via email or LinkedIn.