Biography
I am currently pursuing Computer Science - Master of Science program at Columbia University and possess over 4 years of industry experience in software development. My passion lies in developing practical and innovative solutions to address real-world challenges. I am actively seeking full-time opportunities in Machine Learning Engineering (MLE), Data Science, and Software Development (SDE) roles, starting May 2025.
Interests
- Machine Learning
- Artificial Intelligence
- Natural Language Processing
- Computer Vision
Education
MS in Computer Science, 2025
Columbia University
B.E. in Computer Science, 2019
Netaji Subhas Institute of Technology (NSIT), Delhi
Class 12th, 2015
Indraprastha World School
Experience
Research Assistant
- Research projects focussed on Language Modeling, Conversational AI, and Multimodal Systems under the guidance of Prof. Zhou Yu.
- Developed NLP models using GPT-4 and LLaMA to generate contextual clarifying questions, enhancing comprehension of complex social interactions. Created an abstractive model for QA and employed PPO/DPO techniques to fine-tune models
- Engineered the TLDR Reddit Social Dataset by scraping Reddit posts to build benchmark for question generation task
- Designed and implemented end-to-end pipeline on AWS for question generation, ranking, and benchmarking
Graduate Teaching Assistant (Fall 2024)
- Head Teaching Assistant for COMS 4153: Cloud Computing under Professor Donald Ferguson
- Conducted weekly office hours and training sessions for students
- Graded assignments and projects, while also designing assignment rubrics for all tasks
- Guided and brainstormed with teams on their project ideas, breaking them into multiple sprints
Software Engineer Intern, Intuit Mailchimp
- Worked in Mailchimp SMS team (Delivery squad) on following projects:
- Monolith Proxy Service: Developed a proxy service from scratch to integrate Mailchimp’s SMS stack with monolith application. This service efficiently routes HTTP requests based on tenant information and includes support for caching, error logging, and default fallbacks. Built using Go, Twirp/gRPC, and ProtoBuf, this solution saves developers approximately one hour daily.
- Delivery Re-Architecture: Designed and implemented a revamped architecture with a new data layer, a robust mechanism for delivering messages to downstream aggregators, granular message transition states, and a smart retry system for handling failed or scheduled messages. This re-architecture reduces latency with a lightweight gateway component, increases fault tolerance through the improved retry mechanism, and optimizes database transactions with cleaner data models.
Graduate Teaching Assistant (Spring 2024)
- Head Teaching Assistant for COMS 4111: Introduction to Databases under Professor Donald Ferguson
- Conducted weekly office hours and training sessions for students
- Led a team of 15 TAs, organizing weekly update meetings and delegating responsibilities
- Designed and created weekly assignments and exam papers to enhance course engagement and assessment
Graduate Teaching Assistant (Fall 2023)
- Course Assistant for COMS 6156: Topics in Software Engineering - Cloud Computing under Professor Donald Ferguson
- Conducted weekly office hours and training sessions for students
- Graded assignments and projects, while also designing assignment rubrics for all tasks
- Guided and brainstormed with teams on their project ideas, breaking them into multiple sprints
Senior Software Engineer
- Its Deductible:
- DB Upgrade: Worked on migration of IDO DB from Oracle 12c to 19c and also built a new disaster recovery strategy using AWS Read Replica
- Backend Revamp: Worked on revamp of Its Deductible Backend Application on the latest stack using SpringBoot and Intuit Kubernetes Service
- Smart Data Extraction: Developed a solution using OCR, PII Guardrails and Gen AI(Prompt Engineering) to extract donation details from receipts in strctured format and automatically create donation for users.
- TurboTax:
- Tax Optimization Service - Spousal Support: Worked on adding support for multiple tax payers in optimization problem for TurboTax Canada.
- Global Mobile Tax Platform: Designed solution to decompose logging, analytics and crash monitoring for TurboTax
- TurboTax Mobile: Integrated Firebase Crashlytics and Analytics in TTM application, resulting in early-detection and real time notification of crashes and user app behaviours.
- On-Device Classification: Developed an on-device ML system that accurately classifies tax documents uploaded by users, minimizing the risk of erroneous submissions. The system is optimized through real-time user feedback, resulting in an enhanced user experience.
- TurboTax Orchestration Service: Worked on localisation of content of topics in Tax Orchestration Service and built automated system which saved 50+ hrs during this process.
- Transactions Monolith to Microservice: Contributed to migration of data & building of new APIs with 99% parity
Software Engineer 2
- Mint - Product:
- AffordMint: Developed a model that predicts user's purchase affordability for a category/merchant using transaction benchmarking and various affordability states.
- AI Budget Recommendation System: Built system to predict minters spending and provide insights to create or update budgets, experiment successful with 130 index
- Mint Native to TurboTax Online: Owned integration to help Mint users easily file taxes, generating $6.4M revenue
- Minty Suggestions: Developed iOS widget to suggest top features to the users based on their usage pattern and "Minters Like Me".
- Sweepstakes: Worked end-to-end across Product and Marketing teams to release Sweepstakes in-product campaigns.
- Subscriptions v2: Built a feature for users to manually add and delete their subscriptions, alongwith supporting non-monthly frequencies and getting on-screen feedback.
- Mint - Tech:
- Budgets Widgetization & Re-Arch in iOS: Worked on Budgets iOS widetization, migration from xEvent to GraphQL, and eliminated the persistence store, resulting in a 33% reduction in average API calls and a 7% improvement in build time.
- iOS Design System Library: Developed design system library with 50+ components & integrated with Storybooks
- Mint Life: Started a centralised place and system for documentation of anything and everything required for developers in the Mint.
- Mint Brew: Tool to automate creation of test accounts and generation of insights, helped in saving 200+ hrs of testing.
- App Eye: Built and open-sourced iOS app size scanner to scan iOS app archives and generate treemap/heatmap for file structure.
Software Engineer 1
- Mint - Product:
- Fixed font and color breaking changes on iOS 13 prior to public release.
- Implemented SwiftLint integration to enforce coding guidelines throughout the entire codebase and developed new custom rules to ensure adherence to coding standards.
- AutoMinting: Worked on automating the post-release steps of Mint pipeline.
- Guided Budgets Flow: Developed flow for FTU to add budgets in Mint iOS.
- Mint - Tech:
- Mint iOS Build Platform: Setting automated pipeline to build & deploy the app, increasing release velocity by 2x
- Operational Excellence: Built pod binary caching mechanism to reduce iOS code compile time from 20 mins to 8 mins, reducing web view load time in native apps by 70% and contributing to Mint iOS app size reduction by 27% in 1 week
- Mint De-Junkify: Developed and integrated linting system in Mint iOS that enforced coding guidelines/standards and help reduce code compilation time.
Software Engineer Intern, SRE
- Cloud Raider: Spearheaded the development of a fault injection framework for Automated FMEA (Failure Modes and Effect Analysis), dramatically cutting down the FMEA process from days to hours. The implementation not only enhanced the resilience of over 100 services within the company but was also open-sourced for broader use. The project garnered recognition from senior leadership, and I was honored with an award in acknowledgment of my contributions.
Student Researcher
- Advisors: Prof. M.P.S. Bhatia, Prof. Sushama Nagpal
- Worked on generation of human face images from textual descriptions using ArtistGAN
- Devised algorithm to calculate trust and relationship index in social networks
Technical Content Writer | DS & ML Course
- Created notes for online courses with topics including decision trees, random forest classifier, etc. with practical projects
Teaching Assistant & Problem Setter | DS & Algorithms
- TA to a class of 200+ students for both online and offline courses and curating new problems for online coding platform
Projects
*
Developed a project on controllable movie plot generation from posters using CLIP for image embeddings, a mapping transformer, and GPT-2 for text generation.
Tool to eliminate all the manual pain and effort and completely automate the app size measurement and monitoring in build pipelines. Not currently open-sourced.
Generation of Human Face Images using Text (B.E. Thesis Project). Developed a novel architecture for generation of human face images from textual descriptions. Combination of text encoder, conditioning augmentation, GANs and feature tuning is used to produce high quality images.
Feature Subset Selection
Research project on exploration of feature subset selection using genetic algorithms. Feature subset selection is a NP hard problem and involves selecting of subset of features to increase accuracy and reduce dimensionality. Proposed a wrapper based solution using Vortex Search Algorithm which is a meta-heuristic inspired from vortex pattern created by the vertical flow of the stirred fluids.
Trust in Whatsapp Social Network and Finding most trustworthy group Proposed novel method to find out trust in social network using text, image, audio and video chat communication and calulated relationship index to get the most trustworthy link.
AI Foodie
Food Dishes Recipe Recommendation System. An android application which scans the refrigerator image to recognise the food ingredients using image processing and gives a dish using recipe recommendation algorithm. Not currently open-sourced. Tech - Frontend, Android; Backend - Rails, SQLite; ML - OpenCV, Tensorflow, CNN.
Online Course Management System Platform for students and teachers to facilitate e-learning in colleges. Automates the feedback rating and course selection process for students and uses ML to set personalized assessment for each student based on their performance.
This app automatically posts birthday wishes on your friend’s facebook wall. The birthday wishes written in the message are personalized according to the age of the person.
Volunteering
Computer Science - Department Representative
- Recognized as the Computer Science Department Representative within the Engineering Graduate Student Council (EGSC) at Columbia University's School of Engineering and Applied Science (SEAS)
- Advocated for student interests by collaborating with the administration, while cultivating a sense of community through the organization of events and steadfast support for student club initiatives.
- Worked as a part of Socials team in EGSC and contributed to the organization and coordination of major events such as Rooftop Socials, Carleton Commons, Boat Cruise, and numerous other engaging initiatives.
University Life Campus Ambassador
- Selected as Campus Ambassador for University Life at Columbia University
- Planned and organized Welcome Home Columbia events, including activities like Speed Friending and Silent Disco. Collaborated with a team of 10+ members to successfully execute these events, attracting a total footfall of over 10,000 students.
- Collaborated with the Religious Life team to volunteer for various events, such as MultiFaith Programs, Dinner Events, and more.
- Hosted events on Graduate Students Welcome Day, actively advocating for University Life.
- Awarded the Ambassador of the Month for outstanding dedication, actively participating in nearly every event.
Contact
Thank you for visiting my website. I am currently seeking full-time opportunities in Machine Learning Engineering, Data Science, and Software Development. If my skills and experience align with your organization’s needs, please feel free to contact me!