Mar 2020 – Mar 2021 · 1 yr · Seattle, Washington, United States

May 2019 – Aug 2019 · 3 mos · Gainesville, Florida Area

• Congestion Prediction from Traffic Data using Deep Learning (Python, Pytorch, Folium)
• Transformed the road traffic data into graphs to leverage the interactions between nodes
• Used Graph Convolutional LSTM Network for predicting congestion levels
• Visualized traffic congestion levels and predictions using Folium

Feb 2017 – Jun 2018 · 1 yr 4 mos

• Marketing Campaign Management
• Marketing Campaign Management provides a platform for creation and distribution of user and system generated
• marketing campaigns by leveraging advanced predictive analytics
• Responsibilities
• Developed templates for ad-hoc/scheduled email campaigns using Adobe Campaign. Capable of dynamically reordering the content based on client’s propensity of reading by leveraging the predictive models which drastically increased the click through rate
• Automated the process of content creation using web application in Adobe Campaign
• Made the campaign workflows MiFID and GDPR compliant
• Hadoop Data Integration Utility-
• Created shell scripts to facilitate data transfer (to and fro) between any RDBMS and Hadoop using wrapper shell scripts over Sqoop utility
• Proposed and helped an intern to create a UI based on Angular 2 to manage/schedule/update data transfer

Feb 2016 – Jan 2017 · 11 mos

• Client Interaction Data Delivery
• Client Interaction Data Delivery is a platform to disseminate interactions data to various interaction hubs (OneAccess, CorpAxe, Commcise)
• Responsibilities
• Led the initiative to automate the process, saved ~50 human hours every month.
• Designed and developed data delivery pipelines to Interaction Hubs
• Took the initiative to integrate Support Vector Machine to predict roles as per Interaction Hubs's designated roles from a corpus of 30K roles with an accuracy of ~85% saving the manual effort

Feb 2015 – Jan 2016 · 11 mos

• Unified Pricing Model
• Unified Pricing Model is an outcome of MiFID II regulatory requirement which required Buy-Side Firms to charge fees for the advisory services they offer. The rule based model used resource consumption data to quote a price and catered to both Research and Fixed-Income departments.
• Responsibilities
• Automated the Pricing Model for Research Department
• Architected and developed an application for setting the model inputs, triggering a run and accessing model output in form of one pager summary and consumption reports (DB2, Java, JAX-RS, Spring,Jackson, Freemarker, ExtJS, Highcharts,SAP Business Objects)
• Proposed and implemented the idea to view trends through visualizations to see YoY resource consumption v/s revenue generation for a client
• Data Aggregation in Hadoop-
• Created an aggregated data layer in Hadoop to facilitate running complex queries which were not possible on DB2
• Used shell scripts, autosys and sqoop to automate the process

Aug 2014 – Jan 2015 · 5 mos

• Underwent Technology Analyst Program (TAP)
• Part of the team (3 out of 40 teams) selected for showcasing the TAP project

Jul 2013 – Jun 2014 · 11 mos · Mumbai Area, India

• Worked in the Client Intelligence team which serves the Research Division
• Performed evaluation of Greenplum and Hadoop for heavy computation needs
• Created training dataset [~66 million rows & ~750 features (650 generated)] from user activity data across different channels using Hive
• Worked with data scientists in the team to refine features
• Identified data quality issues to proactively improve model performance
• Optimized the computation performance