Description

Enhanced the backend of a financial analytics platform by integrating machine learning models to assess client creditworthiness and predict the probability of loan default. The platform is used by fintech companies and lending institutions to automate risk assessment and improve decision-making accuracy. My contributions included designing data pipelines for ingesting and preprocessing financial and behavioral data, training and deploying predictive models, and integrating the results into the existing API infrastructure to deliver real-time scoring insights for loan approvals and portfolio management.

Key Contributions

• Advanced AI backend: Rebuilt the backend using FastAPI with advanced data tooling (Pandas, Polars, NumPy, PyTorch).

• Built for scalability: Integrated PostgreSQL, Redis, and async workers for real-time analytics and horizontal scalability.

• Secure AWS hosting: Deployed the full system on AWS, handling VPS provisioning and security configuration.

• Secure Payment Gateway: Integrated a secure payment gateway via Stripe to ensure that all transactions are safe and user data is protected.

Project highlights

• Achieved ~6× faster analytics compared to the legacy system.
• Reduced loan default rates and operational processing costs by ~40%.