May 2024 – Jul 2024 · 2 mos · On-site

• Leveraged a 1 million transaction dataset with Graph Neural Networks to detect money laundering patterns.
• Boosted detection accuracy from 60% to 88% by shifting from traditional models to reduce false alarms for banks.
• Increased the true positive rate by 80% and reduced false positives by 91% and minimizing resource allocation.

May 2023 – Jul 2023 · 2 mos

• Worked on a face verification model for better KYC at banks which solved the problem of verifying old aadhar card photos.
• Made the dataset of 200+ old and new images to test the model accuracy with the Euclidean distance threshold as 0.4.
• Developed MRZ (Machine Readable Zone) text extractor from Td1,2,3 documents using 0CR with decent accuracy.

Mar 2023 – Apr 2023 · 1 mo

• Proposed a blueprint to implement the "Technology Readiness Level" booster scheme aimed at monetarily supporting the
• innovators in enhancing the commercialisation of patented technologies.
• Analysed the market feasibility of a technical invention patented through the Intellectual Property Rights cell IITR .
• Conducted meetings with various business units to collect their insights on the monetary potential of the invention.