Mohammad Affan

AI/ML Engineer with 2+ years of experience building production AI applications, LLM systems and scalable data pipelines using Python, FastAPI, AWS, PySpark, Databricks and MLflow. Skilled in LLM fine-tuning, RAG, NLP and MLOps for reliable production AI solutions, seeking AI/ML opportunities in the UAE and open to relocation with visa sponsorship

AI/ML ENGINEER at Tata Consultancy Services (2024-08 – Present)

• Built Python-based demand forecasting models using XGBoost and Prophet to analyse customer and operational data, achieving 95% daily forecast accuracy to support business planning teams.
• Applied ARDL modelling and Optuna-based hyperparameter optimisation, reducing model search time by 90%.
• Designed and shipped 5+ production AI APIs with Python and FastAPI in a microservices architecture, supporting LLM inference and real-time downstream workloads; optimized serving with vLLM and batching to improve GPU utilization.
• Fine-tuned a domain-adapted LLaMA 3.1 14B model using SFT, PEFT and LoRA for insurance claims recommendations, improving LLM accuracy by 35% and reducing manual effort by 50%.
• Processed approximately 2 TB of data using PySpark and Databricks on AWS, and optimized forecasting workflows with Pandas UDFs and Spark tuning, reducing compute time by 81%; managed 10+ ML/GenAI experiments with MLflow and model versioning.

Bachelor of Engineering (B.Eng.) in Information Technology – Chandigarh University (2020 – 2024)