ML & AI Projects

Seven deployed ML systems across computer vision, NLP, geospatial AI, neuromorphic computing, and MLOps. Each one has a live demo and an open repo.

Neuromorphic Computing

+28 pts under noise

Do Spikes Fail Differently?

A spiking neural network and its ReLU twin, matched in architecture, parameter count, data, seed, and training, compared on DVS128Gesture for accuracy, calibration, robustness under four corruptions, temporal sensitivity, and paper energy. Spiking gives up 3.5 accuracy points and buys a calibration tie, up to +28 points under sensor noise with confidence that tracks its accuracy, and an 18.7x saving on the standard energy accounting. Single seed, n=288, bootstrap CIs on everything, and a live demo.

Natural Language Processing

79.5% win rate

Urdu LLM Fine-Tuning

Qwen 2.5 7B Instruct fine-tuned with QLoRA for Urdu script, Roman Urdu, and Urdu/English code-mixing. The current version wins 79.5% of blind pairwise comparisons against the base model across two independent LLM judges, adds RAG-aware training, and ships as a live Gradio + Modal demo, all for about $60.

Machine Learning & MLOps

73.2% acc · 61% recall

US Visa Approval Prediction

Predicts US PERM labor certification outcomes (73.2% accuracy, 61% denied recall) and explains each prediction with SHAP. A 5-stage MLOps pipeline, GridSearchCV across 5 boosting models, threshold-tuned Gradient Boosting, and a FastAPI backend.

Computer Vision

80% acc · 7 classes

Face Expression Detection

A PyTorch and Flask web app that finds every face in a photo and classifies its expression into one of 7 emotions. ResNet-18 fine-tuned on the RAF-DB dataset, 80% accuracy.

Geospatial AI

R² 0.65→0.89

WaterTrace Pakistan

Groundwater monitoring and forecasting for all 145 districts of Pakistan, built on 22 years of GRACE and GLDAS satellite data (2002-2024). Gradient Boosting time-series model (R²=0.89), interactive district-level maps, and a Flask prediction API.

Web Application

−60% bundle

TeacherRank

A teacher rating and review platform where students score educators on teaching quality, communication, helpfulness, and course difficulty, so others can choose courses with better information. Built with React, TypeScript, and Supabase.

Geospatial AI

90%+ coverage

EV Suitability Analysis

Multi-criteria site selection for EV charging stations in Lahore: weighted scoring over census, OpenStreetMap, and infrastructure data to rank candidate locations across the city's 5 tehsils.