I am a Machine Learning Engineer and Python developer based in Lahore, Pakistan. I build predictive models, automate data workflows, and create interactive dashboards that turn raw data into business decisions.
My technical foundation spans the full data science pipeline — from scraping and cleaning raw data, to training and validating production-ready ML models, to presenting results in Streamlit dashboards and automated Excel reports.
What I offer:
• Predictive modeling with LightGBM, scikit-learn, and RandomForest (regression and classification)
• Honest model validation using GroupKFold cross-validation — results that hold on real unseen data, not just training data
• Feature engineering for tabular and sequential (time series) data
• Web scraping, REST API consumption, and data pipeline automation using Python
• Interactive dashboards built with Streamlit, Plotly, and pandas
• Excel and CSV report generation with openpyxl
Why I am different from most ML freelancers:
I do not just fit a model and show a training accuracy number. Every model I deliver is validated honestly — held-out folds, no data leakage, real-world RMSE. I learned this the hard way competing in a live $50,000 Kaggle competition (ROGII Wellbore Geology Prediction) where I currently rank in the top cohort after implementing rigorous local validation before any submission.
Projects I have built:
• Sales Analytics Dashboard — CSV upload, 4 KPIs, 3 charts, Excel export (Streamlit + Plotly)
• Blockchain Analytics Dashboard — live CoinGecko API, Bitcoin 30-day chart, sidebar filters
• Customer Churn Predictor — RandomForest, risk scoring, feature importance (scikit-learn)
• Kaggle competition pipeline — LightGBM with delta target framing, GroupKFold, 61-feature engineering on 773-well industrial dataset
Ideal projects:
Regression and classification problems, sensor or log data analysis, automation scripts, dashboards with live data, custom ML pipelines, and data cleaning workflows.
I respond within 24 hours and provide clean, documented code with every delivery.