Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I'm Mathai Sibu Kombasseril, a Data Scientist specialising in predictive modelling, optimisation, and production-grade ML systems for risk, revenue, fraud, and operational performance. I design and deploy end-to-end ML and GenAI workflows across lending, financial services, and high-volume environments to turn data into measurable business value. I have a proven track record of delivering measurable commercial impact—8% uplift in approvals, 50% fraud-loss reduction, and 40% cost reductions through automated decisioning and GenAI pipelines. I excel at building audit-ready MLOps, integrating ML outputs into enterprise systems, and collaborating with engineering, product, and leadership to scale adoption and reliability.…I'm Mathai Sibu Kombasseril, a Data Scientist specialising in predictive modelling, optimisation, and production-grade ML systems for risk, revenue, fraud, and operational performance. I design and deploy end-to-end ML and GenAI workflows across lending, financial services, and high-volume environments to turn data into measurable business value. I have a proven track record of delivering measurable commercial impact—8% uplift in approvals, 50% fraud-loss reduction, and 40% cost reductions through automated decisioning and GenAI pipelines. I excel at building audit-ready MLOps, integrating ML outputs into enterprise systems, and collaborating with engineering, product, and leadership to scale adoption and reliability.

Mathai Sibu Kombasseril

Data Scientist, AI Engineer, Data Analyst, +3





I'm Mathai Sibu Kombasseril, a Data Scientist specialising in predictive modelling, optimisation, and production-grade ML systems for risk, revenue, fraud, and operational performance. I design and deploy end-to-end ML and GenAI workflows across lending, financial services, and high-volume environments to turn data into measurable business value. I have a proven track record of delivering measurable commercial impact—8% uplift in approvals, 50% fraud-loss reduction, and 40% cost reductions through automated decisioning and GenAI pipelines. I excel at building audit-ready MLOps, integrating ML outputs into enterprise systems, and collaborating with engineering, product, and leadership to scale adoption and reliability.…I'm Mathai Sibu Kombasseril, a Data Scientist specialising in predictive modelling, optimisation, and production-grade ML systems for risk, revenue, fraud, and operational performance. I design and deploy end-to-end ML and GenAI workflows across lending, financial services, and high-volume environments to turn data into measurable business value. I have a proven track record of delivering measurable commercial impact—8% uplift in approvals, 50% fraud-loss reduction, and 40% cost reductions through automated decisioning and GenAI pipelines. I excel at building audit-ready MLOps, integrating ML outputs into enterprise systems, and collaborating with engineering, product, and leadership to scale adoption and reliability.

Available to hire

I’m Mathai Sibu Kombasseril, a Data Scientist specialising in predictive modelling, optimisation, and production-grade ML systems for risk, revenue, fraud, and operational performance. I design and deploy end-to-end ML and GenAI workflows across lending, financial services, and high-volume environments to turn data into measurable business value.

I have a proven track record of delivering measurable commercial impact—8% uplift in approvals, 50% fraud-loss reduction, and 40% cost reductions through automated decisioning and GenAI pipelines. I excel at building audit-ready MLOps, integrating ML outputs into enterprise systems, and collaborating with engineering, product, and leadership to scale adoption and reliability.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Language

English

Fluent

Work Experience

Senior Data Scientist at Around Zero Ltd

March 1, 2022 - Present

Led GenAI & risk automation initiatives, architecting production-grade Retrieval-Augmented Generation pipelines using LLMs to automate complex document intelligence, reducing manual processing costs by 40% and unlocking actionable decision insights from unstructured data. Designed scalable ML infra on AWS (S3, Lambda, SageMaker) with CI/CD and MLflow, boosting deployment velocity by ~50% and delivering real-time scoring. Transformed manual risk assessments into automated ML-driven scorecards, achieving 8–12 pp KS lift and 5–8% reduction in preventable losses. Built real-time fraud detection reducing false positives to 1.5–2.5% and halving fraud losses. Implemented SHAP/LIME explainability and audit-ready governance, ensuring compliance and facilitating executive decision-making.

Principal Logistics Analyst at Grace Rides Ltd

August 1, 2018 - February 1, 2022

Developed ML-driven demand forecasting models to optimize regional inventory, cutting shortages by 28% and delivery costs by 12%. Implemented a fraud prevention system saving £850K annually with 1.8% false positives. Leveraged NLP for inquiry routing in operations, automating customer queries and reducing manual handling time.

Operations & Inventory Analyst at Tesco

October 1, 2006 - July 1, 2018

Replenishment modelling reduced forecast error from 20% to 10% and stockouts by 25%. Conducted promotional analytics with statistical modelling and A/B testing to measure lift, achieving 95%+ service level on high-volume retail portfolios.

Founder & Senior Data Scientist – Risk, Revenue & GenAI at Around Zero Ltd

March 1, 2022 - Present

Built approval-risk models delivering 8% approval uplift and 40% default reduction, forming early-warning indicators for portfolio-level risk. Delivered RAG/LLM document-intelligence workflows achieving 40% cost reduction and 90% automation. Owned full production lifecycle for ML and GenAI systems — engineering datasets, building models, deploying to cloud, implementing monitoring, and maintaining live scoring pipelines used in daily decisioning. Re-engineered fraud-signal pipelines to achieve 50% fraud-loss reduction with 2.5% false positives. Improved capital recovery by 12% and reduced churn by 20% using behavioural segmentation. Cut deployment timelines by 40–50% through CI/CD-aligned governance and automated monitoring. Engineered large-scale datasets using Python, SQL and PySpark; led cross-functional ML/GenAI solution design; integrated outputs into Power BI dashboards for programme delivery risk modelling.

Founder & Data Scientist – Forecasting, Optimisation & Revenue Systems at Grace Rides Ltd

August 1, 2018 - February 1, 2022

Delivered 60% conversion uplift and 20%+ acceptance improvement via segmentation and propensity modelling. Prevented £850K annual leakage with fraud controls at 1.8% false positives. Reduced inventory shortages by 28% through ML forecasting (30% MAPE improvement). Cut marketing cost by 20% and increased sales productivity by 25% via automated arbitration logic. Built NLP-driven classification and routing systems reducing manual handling time. Developed scalable pipelines enabling continuous model updates, drift detection, and operational monitoring. Built and maintained production ML pipelines powering real-time segmentation, fraud detection, and forecasting, with automated updates, monitoring, and continuous optimisation.

Data Analyst & Team Leader – Stock, Demand & Inventory Analytics at Tesco

October 1, 2007 - July 1, 2018

Developed analytical systems modernising replenishment, inventory, and promotional decisioning across large-scale retail operations. Improved working-capital efficiency by 12% through automated fund allocation and optimisation. Maintained 95%+ service levels while protecting margins using uplift modelling and experimentation. Implemented production analytics and forecasting systems embedded into enterprise replenishment and inventory workflows, enabling automated decisioning at national scale. Cut forecast error by 50% and automated 50% of manual replenishment interventions. Supported enterprise retail operations with forecasting, analytics, and decision-support frameworks.