M

M-Musa Khan

Kaggle Master and AI/ML Engineer specializing in Computer Vision and Predictive Analytics.

Based in: Lahore, PakistanMost recently: Data Scientist(AI/ML Engineer), WaseelaExperience: 4 years

A highly skilled Data Scientist with a proven track record in deploying high-accuracy CV systems and hybrid forecasting models that significantly reduce operational costs. He is a global top 1% Kaggle Master currently pursuing a BS in Computer Science at UET Lahore.

Experience

Data Scientist(AI/ML Engineer)

WaseelaPakistanworkJul 2025 – Apr 2026

Key Responsibilities: • AI Customer Counter: Engineered a CV system to transform standard security cameras into automated customer counters, optimizing a YOLO BOT-SORT model for low end systems using existing camera infrastructure on real-time CPU performance. • AI Demand Forecasting: Architected a hybrid model (Linear Regression, Catboost, TimeGPT) that slashed forecast error (WAPE) from 400% to 20-50% for volatile products, deployed via a Streamlit dashboard. • Logistics Route Optimization: Built an automated engine to solve complex Vehicle Routing Problems (VRP) using Google OR-Tools and OpenStreetMap, significantly reducing daily logistics costs. • Field Segmentation: Enabled live agricultural land mapping by deploying a SAM2 powered field boundary segmentation model, achieving 98% boundary extraction accuracy while dynamically processing satellite imagery through tunneling to production. The system was able to process upto 1.8 million marlas of land in a single batch. Achievements: • Slashed forecast error (WAPE) from 400% to 20-50% for volatile products. • Achieved 98% boundary extraction accuracy for agricultural land mapping. • Processed up to 1.8 million marlas of land in a single batch.

YOLOBOT-SORTLinear RegressionCatboostTimeGPTStreamlitGoogle OR-ToolsOpenStreetMapSAM2Computer Vision

Data Scientist

Brokers LoungeRemotefreelanceOct 2022 – May 2025

Summary

Key Responsibilities: • Built automated Python-based web-scraping and ETL pipelines to extract property data from multiple websites, transforming fragmented market information. • Designed comprehensive dashboards to visualize investment hotspots, providing brokers with clear, data driven insights to guide profitable real estate investments.

Results

Consolidated property data into a visual map representation to highlight the best opportunities.
Scraped approximately 10,000 properties per data source.
Saved brokers hours and days of manual work by consolidating property data into dashboards.

PythonWeb-scrapingETLDashboardsSQL-lite