Kshitij Mathur

- Developing Computer Vision models and deploying them on AWS using SQS, S3, EC2

- Part of the Product Development Division responsible for providing daily order proposals for 13 retail customers across 7 timezones. - Designed and Implemented an Ingestion pipeline to ingest ML predictions using pre-caching techniques to achieve an API response time <3 seconds. Integrated ML Algorithms with existing UIs to allow for manual override of ML Predictions - Parallelized Machine Learning algorithms and reporting tools using Dask Distributed Computing with 750+ partitions for 500 million records daily. - Responsible for major design-level decisions concerning improvements and technical debt in data ingestion and UI functionality. - Re-designed entire backend service architecture to decouple from other products/services and achieve a 30% reduction in the code base, 2x improvement in CI pipeline runtimes, and 2x efficiency of the team. - Implemented clustering logic to improve ML pipeline run-time by 30%. - Led the CI/CD migration from InHouse Jenkins to Azure-Based Service. - Took 100+ interviews for the role of Junior Data Analyst, Data Scientist, Junior Software Engineer, and Senior Software Engineer

- Built an end-to-end data science pipeline reducing time spent per RFP by more than 60% - Developed a combination of heuristics and machine learning techniques (~85% recall) to extract questions and answers from information from documents in PDF, Doc, HTML, and Docx formats. - Developed unsupervised keyword extraction using a research paper - Developed an algorithm to return similar questions for a given question. - Developed metrics and visualizations using transportation route and ticket data to identify bottlenecks and inefficiencies in the transportation system.