May 2015 – Feb 2025 · 9 yrs 9 mos · Bangalore · Hybrid

• Currently leading a team of 10 MLEs with focus on creating end to end ML architecture for machine learning within Compliance Engineering. The role entails identifying the technologies to use for feature engineering, the data storage format and schema and applying advanced ML techniques to build a high precision/ high recall surveillance portfolio for Compliance Engineering.
• Across my tenure in Goldman Sachs I have been involved in architecting big data surveillances in the Compliance Engineering domain including Insider Trading, Entity Recognition, Insider Threat surveillances.My role currently entails building large-scale ML models which use distributed data processing frameworks like Scala, PySpark, Map Reduce for feature generation and advanced ML techniques like Neural Networks, Large Language Models to create efficient models for Compliance.

Apr 2014 – May 2014 · 1 mo · Bangalore

• Improving the Capture Rate and Coverage of existing risk models for capturing people who default with high balances on their credit cards by introducing incremental variables and evaluating automated techniques like Gradient Boosting Methods(GBM).

Jul 2012 – Jun 2013 · 11 mos · Hyderabad Area, India

• At Qualcomm, I have been working on the power optimizations for the current as well as upcoming Wireless Solutions. I have been involved in the bring up and commercialization of various chips and have worked independently on numerous optimizations.

May 2011 – Jul 2011 · 2 mos · Augsburg, Germany

• The project involved the use of Ray features and Graph Cut Algorithm to devise an optimum technique for Object Recognition of a wide variety of objects we encounter in our day to day lives. This is a novel method in the sense that the Ray features were currently used only for object segmentation in molecular biology.

Jun 2010 – Jul 2010 · 1 mo · South Korea

• The project involved using SIFT features to capture significant traits for object segmentation using advanced computer vision algorithms