I build machine learning solutions, fraud detection systems and data-driven applications — from experimentation and modelling to APIs and real-world deployment.
Experience across banking, fraud analytics, machine learning, model development and real-time decision systems.
Working on fraud detection models, real-time scoring, model implementation, feature engineering and analytics for large-scale banking systems.
Worked on data science, analytics and machine learning solutions for banking use cases with focus on turning data into actionable business decisions.
Selected projects across machine learning, fraud analytics, API engineering and generative AI.
End-to-end transaction fraud modelling pipeline using historical behavioural features, model validation, feature selection and fraud capture analysis.
Designed and optimized model scoring workflows with focus on low latency, reusable feature generation and scalable real-time model execution.
Retrieval-augmented generation application supporting document ingestion, embeddings, vector search and contextual question answering.
Automation framework for generating and executing high-volume model test scenarios to improve validation coverage and reduce manual testing.
Lightweight ARM API server running multiple independent applications using Android, Termux, Python and FastAPI.
This portfolio is also my playground for experimenting with machine learning services, APIs and data products.
Experimental APIs and applications hosted on my personal self-hosted infrastructure. Availability may vary while services are under development.
Machine learning endpoint for evaluating transaction fraud risk.
General-purpose FastAPI service hosted from my self-managed server environment.
Technologies I use across data science, machine learning, analytics and application development.
Interested in data science, machine learning, fraud analytics and building useful technology.