- Designed and developed the C3 data lake for storing and maintaining AIS data of the vessels traveling across the world in Postgres and Cassandra tables, queried by the supply chain applications for predicting shipment delays.- Designed and developed the Bill of Materials graph feature of the Inventory Optimization (IO) application, which helped the customer visualize and predict the facilities affected by shortage of an item at a given facility.- Enhanced the IO application scalability and performance by 70%, by revising the UI configuration workflow, reducing the back-end API calls and caching instances.Implemented the feature metrics of IO, fed to the ML pipelines for predicting new inventory replenishment numbers, delivery date of a purchase order line.- Built custom and reusable data grid component using React, which facilitated inline cell editing for the company’s component library.

- Developed the Demand Forecast Subject details page of Demand Forecasting application, involving components like metric tiles for subject’s KPIs, interactive grid for its forecasts using React, Redux and C3 AI component library.- Migrated the ExternalDataFeeds package and its dependent packages (InsideView & Quandl ) from C3 V7 server version to V8 including fixing pre-existing test scripts migrated by the migrator tool, achieving CI/CD using Jenkins.

- Designed and developed an end to end code recommendation engine to recommend pre-existing C3 Types and APIs to developers implementing new APIs in JavaScript in less than 5 seconds.- Implemented a customized clustering, and pruning algorithm using JavaScript, to cluster all the code snippets from C3 repositories, based on syntactic and tree structures, and prune them w.r.t the query snippet provided by developer seeking service.

SLATS Team- Data Analyst

Creating Assignments and Presentations for CSCI 570 - Analysis of Algorithms

- Utilized LSTM and RNN models for automating complaints classification of a customer query ticket management system which increased its efficiency by 80%, and created routes to integrate it to the system using Django.- Developed a KYC documents image classifier using complex CNN model for 1000+ unstructured documents distributed over 30 classes, attaining over 95% accuracy and yielding upto 70% reduction of work time per document.- Developed a decision-support system for a well known Britain retailer, developed a Random Forest Regression model with WAPE ~ 13% to strategize product launching decisions over 20 categories.