|B.Tech in CSE||National Institute of Technology, Dgp||8.46||2013-17|
|Online Masters in Computer Science||University of Texas, Austin||3.67||2020 - 2021|
|Masters in Computer Science||University of California, Irvine||2021 - current|
INTUIT FUTURES (R&D) SOFTWARE ENGINEER 2 (Feb 2019 - July 2021)
Federated Learning Platform:
Ideated and Developed an FL platform to train Machine Learning models. Enabled Intelligent Use Cases for our 2 million QuickBooks Desktop customers, since direct data access is restricted, prohibiting traditional approaches.
Leveraged this platform to train ML model. Predicted Invoice Payment dates to help customers with better cash flow planning.
Cerebro: Provides real time insights about Intuit’s competitor products using public and social data. Pivotal in $7.1 billion acquisition of Credit Karma by Intuit, multiple aquihires and tracks market statistics.
Owned unsupervised KeyPhrase Extraction model and Sentiment Analysis Model. Saved users time spent on manually going through all reviews by providing weekly user reaction summary about competitor products.
Led design of intelligent insights (Anomaly Detection, trends in app update vs ratings) from LinkedIn, Glass Door and App-Annie by building AWS server-less infrastructure and D3.js. It handles 3.5GB every month.
Quickbooks Self-Help: Conversational User Interface. Answers user queries based on self-help articles and highlights most relevant paragraph within document.
Built information retrieval and information extraction model. Retrieved document, matching user query with BERT Em- bedding and highlighted correct paragraph using Bag-of-Words. Increased usage of self-help by 35% and decreased care calls by 66%.
QUICKBOOKS SOFTWARE ENGINEER (Aug 2017 - Jan 2019)
Digital Tax Experience for UK: Developed Digital Tax experience for Quickbooks UK that helped 165k customers file their taxes.
MCS: Infrastructure for storing application configuration to make it variable across different dimensions of product.
Constructed a Recommender System to suggest products to consumers based on products sold by similar retailers based on Graph database (Neo4j). Reduced time taken to enter new products to seconds.
RESEARCH PROJECTS & PUBLICATIONS
Designed a variation of K Nearest Neighbours algorithm which achieved an accuracy of 98.83, beating previous high score on UCI Wisconsin breast cancer data which was 98.61. Published in Asian journal of Applied Sciences.
Designed an ensemble model to filter new updates and bugs from user reviews. Published in ICMLSC 2020.
SuperTuxKart Ice Hockey Player (UTAustin) - Designed a striker and a defender using Reinforcement Learning (DQN) to drive and CenterNet based model to detect Puck. We ranked 4th in final tournament.
Contributed to Scikit-learn open-source library by writing Invariance Tests for Clustering Metrics.
SPEAKING AND WRITING OPPORTUNITIES
Presented at Intuit Internal events on Deep Learning, Federated Learning and Natural Language Processing.
An ardent blogger - Medium.com
Written articles for Intuit in DQIndia and NASSCOM community.
SOCIETY MEMBERSHIPS: WOMAN IN TECH
Mentored woman undergraduates as part of woman’s development team in Intuit.
Given talks and hosted events as part of tech woman at Intuit community.
Part of university recruitment in Intuit.