SURABHI S

Data science master's student with over six years of experience in data-focused roles, currently serving as a machine learning engineer since April 2024. Expertise in developing and deploying machine learning and predictive models, including large language models and CI/CD pipelines. Proficient in Microsoft Fabric modern MLOps practices and end-to-end model lifecycle management. Strong background in data collection, processing, and analysis to drive data-driven decision-making.

• Mitie Projects: Led LLM-driven categorisation projects at Mitie, processing large datasets using async pipelines to optimise throughput and run tasks in efficient batches. Integrated GPT models via the OpenAI API in Microsoft Fabric, implementing robust pipelines with code review, monitoring, and quality controls.
• Designed reusable, no-code LLM summarisation templates that allow non-technical users to generate consistent, high-quality summaries without needing coding expertise. Standardised inputs, prompts, and workflow steps to ensure ease of use and reliability across teams.
• Developed Salesforce predictive models using XGBoost, focusing on feature engineering, model optimisation, and business KPI alignment. Improved model performance through hyperparameter tuning with Optuna, resulting in more accurate and reliable predictions.
• Masters: I have hands-on experience in data analysis and machine learning through a significant project where I analyzed open accident data (2010-2020) from gov.uk. In this project, I not only performed data cleaning, exploratory data analysis, and feature engineering but also effectively addressed the high imbalance in the data. Handling this data imbalance was crucial for achieving accurate results in the machine learning model. The model, which predicted the criticality of accidents based on factors such as vehicle type and geographical location.

• Microsoft Certified: DP-600 – Implementing Analytics Solutions Using Microsoft Fabric (2025)
• Data Science Foundations - Level 1 – Issued by IBM
• Deep Learning Essentials – Issued by IBM
• IBM Agile Explorer