Anurag Tangri

About Anurag Tangri

Working as Principal Applied AI Scientist @ Risk & Identity team with the focus on…

Experience

1. Lead Ml Scientist, Visa Research

   Visa

   Jan 2019 — Present

   Working as Lead ML Scientist with focus on applying AI and ML techniques: A. Real Time Payments (RTP)(Oct 2020 - current)- Created an unsupervised Deep Learning Mule Detection model based on Variational Auto-Encoder (VAE) to identify mules and associated collusive communities- Graph Representation Learning to create a novel set of Graph Features that have been a major driver to improved model performance in real-time fraud detection. These features are first ever to be deployed and used in Visa at scale for predictive modeling- Multiple client POCs all around the world to test and enhance real-time Transaction Scoring model on real RTP client data- 4 Patents Filed B. Merchant Entity Resolution (Dec 2019 - Sep 2020)- Created a Merchant Entity Resolution system to identify merchants changing their attributes in Visa’s payment system. Able to capture 80% of the disappeared merchants on daily basis with high accuracy- The model uses an ensemble of Feature Similarity algorithm based on Apache’s Spark Local Sensitive Hashing and a consumer behavior approach based on Jaccard Similarity- The model was also selected for a Tech Talk at Spark AI Summit 2020: https://databricks.com/session_na20/using-ai-to-support-proliferating-merchant-changes - 3 Technical Innovations awarded. C. Visa Business Solution DQ Rules Engine (July 2019 - Nov 2019)- Created an AI based rules engine to identify rules based on the quality of data sent by merchants- The engine identifies a set of rules based on type of data, numerical relationships, uniqueness of fields and NLP and is used by Acquirer banks to identify merchants sending incomplete data in transactions. D. Visa Strategy Manager (Jan 2019 - June 2019)- Enhanced Random Forest Apache Spark based rule recommendation engine to create and recommend fraud capturing rules for Banks that issue cards to people- Significant revenue making service marketed to 140 banks bringing huge revenue to Visa- 1 patent filed

Skills

• Hadoop
• Hive
• Distributed Systems
• Unix
• Software Development
• Perl
• Data Structures
• Eclipse
• Xml
• C++
• Solaris
• Multithreading
• Cdh
• Python
• C
• Apis
• Linux
• Subversion
• High Performance Computing
• Clearcase
• Java