Machine learning and Python developer. I build end-to-end ML pipelines, NLP models, and AI agents — from autonomous phishing detection agents with tool-calling and progressive skill disclosure, to fine-tuned sentiment models for Roman Urdu.
| Project | Description | Tech |
|---|---|---|
| Phishing Email Detection Agent | Autonomous tool-calling agent that detects phishing emails using progressive skill disclosure, VirusTotal URL scanning, WHOIS domain checking, and prompt injection screening. 99% accuracy on Nazario + Enron dataset. Built for Google x Kaggle 5-Day AI Agents Intensive. | Python, Flask, Groq, Llama 3.3 70B, VirusTotal API, Chrome Extension |
| Roman Urdu Sentiment Bot | Sentiment analysis chatbot for Roman Urdu YouTube comments | DistilBERT, Streamlit, PyTorch |
| Parkinson's Disease Detection | ML pipeline across 3 UCI datasets with cross-dataset generalization testing | scikit-learn, Python |
| Library Management System | Terminal-based library system with OOP and JSON persistence | Python |
| Cafeteria Pay | Real-time payment queue simulation with producer-consumer threading, semaphores, and priority scheduling | Python |
| Python Mini Projects | ATM simulation, number guessing game, to-do list manager | Python |
| ML Bench Internship | Daily ML projects covering data structures, preprocessing, regression, classification, and data analysis | Python, scikit-learn, pandas, NumPy |
Python, scikit-learn, PyTorch, HuggingFace Transformers, Streamlit, Flask, Groq API, Llama 3.3 70B, VirusTotal API, pandas, NumPy, threading
- GitHub: @hifsaiftikhar
- LinkedIn: Hifsa Iftikhar