Jun 2021 – Jul 2024 · 3 yrs 1 mo · Bengaluru, Karnataka, India

• Worked with consumer science team primarily in the risk control and recommendation domain responsible for design, development and deployment of solutions along with governance and monitoring.
• Worked extensively on operational risk models across multiple entities (merchants, customers and drivers) and services for Gojek in multiple geographies. Implemented graph ML solutions on homo/heterogeneous graphs to curb collusion fraud involving diverse modus operandi. Experienced in deploying scalable solutions in batch, near real-time and real-time fashion.
• Collaborated with search and relevance team to implement Graph Neural Network (GNN) based recommendations for GoFood app with a focus to enhance customer experience, discovery and optimising merchant impressions. Involved in end to end GNN model training and designing scalable data pipeline to process large scale transaction graph followed by subsequent online experimentation.

Jun 2020 – Jun 2021 · 1 yr

• I developed model monitoring and governance framework on Hunter (Experian’s fraud engine) platform. I work closely with Experian XLab to implement online/active learning framework on Hunter fraud score for fast and efficient capture of dynamic fraud patterns. The entire solution was selected as runner-up for Experian’s Si Ramo for commercial Impact 2020.
• I worked with alternate data sources implementing text mining techniques on unstructured data for better profiling of NTC (new to credit) customers, as a part of Experian's Prism initiative to use consumer consented data for credit access.

Jul 2018 – Jun 2020 · 1 yr 11 mos

• I collaborated with Experian XLab’s open banking problems using transaction data. My contribution lies in the development of semi-supervised transaction representations and transformer based recommendation system for offers. This project was the runner-up for Experian’s Si Ramo for Scientific Excellence 2019.
• I worked on Hunter (Experian’s fraud engine) platform to implement ML - applications. In particular, I developed and implemented the Hunter fraud score for faster application processing. My experience in Hunter fraud score extends to APAC countries as well.

May 2017 – Jul 2017 · 2 mos · Greater Hyderabad Area

• I worked as a Machine Learning intern to develop information extraction system from unstructured job description.
• My work was mainly in the area of syntactic and dependency parsing to extract relevant information chunks using OpenNlp and configure the system as per REST API for deployment.