Rigved Manoj

Machine Learning Intern

Changing Hands•  June 2024 - August 2024•  Remote, USA

Developed a pricing model for second-hand fashion utilizing a hybrid regression approach that integrates BERT for text analysis and a neural network to handle categorical features. Compiled a dataset of 1 million clothing items by scraping online platforms with Beautiful Soup library and applied advanced feature engineering with Pandas and Numpy to extract valuable features. Utilized a custom loss function (mean squared percentage error) to enhance the model’s accuracy, achieving price predictions within a 30% range of actual values.

Senior Software Engineer

Visa Inc•  March 2023 - July 2023

Handled 15 million daily transactions by crafting a resilient data ingestion system leveraging HDFS, Spark, and Python for distributed processing, and implemented CI/CD pipelines to automate testing and deployment, ensuring seamless integration. Reduced code redundancy by 43% and lowered resource utilization by 10% by building a high-performance ETL tool using Python and Hive, automating data extraction and transformation across applications. Identified and resolved a critical data discrepancy issue by conducting root cause analysis, exploring data lineage, and conducting thorough code review, resulting in preserving $240k in revenue.

Software Engineer

Visa Inc•  October 2020 - February 2023

Enhanced an internal tool using React and Bootstrap to display job failures over time via interactive dashboards, facilitating teams in monitoring and examining failure trends. Redesigned applications to align with India’s new central bank security guidelines, ensuring 100% compliance and data localization, while minimizing disruptions for end users and clients. Remediated 3 years of data inconsistencies involving 10 billion records by scrutinizing data lineage, executing incremental data backfilling, and consistently overseeing systems to guarantee seamless service for end users. Implemented a scalable, configuration-driven tool in Python to automate data pipeline testing, resulting in a 32% decrease in manual testing workload through integrated parallel running.

Machine Learning Intern

Visa Inc•  May 2019 - July 2019

Created a Python-based full-stack application to predict transaction patterns utilizing Meta’s Prophet model for anomaly identification, leading to an 18% enhancement in accuracy compared to the prior model. Improved recall and enhanced precision by 25% and 10%, respectively, utilizing a structured hyperparameter tuning strategy that integrated grid search and random search techniques.