Parth Agarwal

Initiated and delivered a Liveness Detection product from concept to execution, distinguishing live persons from presentation attacks and handling ~3M API calls/month, preventing fraud.

Achieved performance surpassing competitors' existing products within 1 year by improving model accuracy by ~12% and reducing False Rejection Rate by >40% through advanced data generation and model optimization.

Headed the MLOps initiative, deploying MLflow for experiment tracking and data annotation, streamlining workflows for 20+ Data Scientists, 6 Data Annotators, and Product Managers.

Authored 2 packages to standardize MLflow usage and streamline regression testing for Data Science product APIs, enhancing codebase efficiency.

Engineered a Video Liveness product from scratch, incorporating 25+ API calls and creating scoring mechanisms, reducing dependency on vendor products and cutting costs by ~45%.

Developed a generic LLM platform that reduced Turnaround Time (TAT) for new product requests from ~2 weeks to 1 day and cut costs by ~90%.

Led a 6-person team, establishing the Data Science team and overseeing multiple projects including Face Verification, Face Detection, and Image Quality Assessment.

Designed and created automated ETL pipelines, integrating a variety of raw data sources and optimizing for OLAP design requirements.

Successfully implemented a robust ETL pipeline to migrate critical client data from existing RDBMS to AWS Redshift Cluster, facilitating OLAP creation.

Leveraged a suite of AWS services including S3, Lambda, and Redshift, alongside Apache NiFi, to ensure seamless and efficient data migration processes.

Completed intensive 6-month training in Software Development Engineering (SDE) and advanced Big Data concepts, enhancing foundational technical expertise.