M Ikram Ullah

Data Scientist | Researcher

University • Parttime

Jan 2025 - Present • 1 yr 4 mos

Provide end-to-end Data Science and AI solutions for diverse clients, specializing in predictive modeling, NLP, and computer vision research. I bridge the gap between academic research and industrial application by deploying explainable, high-performance ML systems. Core Freelance Services & Achievements: Custom ML Pipelines: Architected and deployed 6-step machine learning pipelines, including a telecom churn prediction system achieving 82.1% accuracy and 0.847 AUC-ROC. Deep Learning & Computer Vision Research: Conducted advanced research on Vision Transformers (ViT) and CNNs, implementing custom loss functions (Focal Loss) to solve extreme class imbalance in facial emotion recognition tasks. Predictive Analytics & Fraud Detection: Built robust fraud detection systems utilizing XGBoost and LightGBM, handling imbalanced datasets with SMOTE to achieve 0.95+ AUC-ROC. Explainable AI (XAI): Integrated SHAP and LIME into client models to provide feature-level transparency, ensuring models meet compliance and audit requirements. Natural Language Processing: Developed real-time sentiment analysis tools and NLP-driven dashboards featuring custom lexicon engines and negation detection. Automated Data Solutions: Created AutoEDA tools to generate automated statistical summaries, outlier detection, and data quality reports, streamlining the discovery phase for client projects. Time Series Forecasting: Leveraged ARIMA, Prophet, and LSTM models to deliver multi-step energy consumption forecasts based on historical seasonal data. Technical Expertise Offered: Languages & Frameworks: Python (Pandas, Scikit-learn, PyTorch, TensorFlow), SQL. Specialized Research: Class Imbalance Handling, Transfer Learning, and Statistical Modeling.

AI/ML Engineer

Syslab Technologies • Fulltime

Sep 2024 - Dec 2025 • 1 yr 3 mos

Making intelligent systems with Ai integration. Cross platform apps for both iOS and Android.