• Built a distributed, multi-region ML ETL orchestration framework using SNS/SQS fan-out scheduling and Lambda state-based workflow

orchestration, automating EMR and SageMaker jobs with automatic retries, idempotent execution, and cross-region VPC networking to

deliver a fault-tolerant, highly-available pipeline.

• Designed reusable backend microservices and APIs, enabling downstream teams to integrate with our system and allowing our pipeline to

consume upstream team services with minimal onboarding effort.

• Implemented infrastructure-as-code using AWS CDK, improving deployment consistency.

• Added dashboards, alarms, and automated ticketing using CloudWatch, improving system observability and reducing on-call noise.

• Modernized legacy Sponsored Brand Ads services, reducing maintenance effort across internal ad platforms.

• Developed new UI cards and a recommendations page in Sponsored Brands’ Campaign Builder (TypeScript/JavaScript), surfacing

ML-generated campaign recommendations from our ETL pipeline and enabling broader advertiser adoption of automated campaign setup.

• Transitioned infrastructure to a serverless micro-service architecture using ECS Fargate, ECR, Lambda, and Docker

• Implemented CI/CD for ElasticBeanstalk using CodePipelines/Github and for ECS/ECR using Github Workflows

• Fullstack development with ReactJS frontend and Flask backend

• Implemented best practices for security using IAM

• Made AWS infrastructure highly available and scalable using load balancing and auto scaling

• Implemented metrics and alarms using AWS Cloudwatch

• Created and maintained EC2 servers

• Implemented closed captions support on Sponsored Brand Video Ads on multiple video ad widgets

• Implemented frontend framework for captions as well as a captions button

• Implemented backend metadata parsing

• Gated all changes behind a Weblab

• Implemented metrics to track, analyze, and monitor the use of these features

• Used AWS services such as S3, EC2, Athena, CloudSearch, and Lambda

• Used Git for version control

• Performed necessary testing and debugging in pre-prod/gamma stage

• Used mock requests and responses to test end to end throughout all services

• Fullstack development in HTML/Kata, Typescript, and Java

• Worked on creating the Hardhat development environment for the Zap platform with a focus on oracle services.

• Helped to develop and stress test a miner for mining zap tokens

• Worked on the Convective Cores project focusing on using unsupervised clustering to automatically label which convective core a particular pixel belongs to and then track them through time. Once labeled we determined the average statistics of these cores and the sensitivity of the results in regard to our algorithm.

• Mentored younger students helping them with python skills as well as plans to approach tasks and how to organize team dynamics for group projects.

• Researched the role of clouds in weather and climate through computational analysis of very large datasets of cloud properties collected by national research institutions such as NOAA and NASA.

• Coded in python to create models from big datasets to show the discrepancies between ground cloud observers and satellite cloud observations.

• Studied Polar Lows in the Arctic and Antarctic regions and helped to develop a database of polar lows to be used by a machine learning algorithm to identify polar lows using object recognition and computer vision.

• R Singh, E Schneider, A Friedman, R Panickar. Investigating Convective Cores in Time and Space. Presented at: FIRE Summit. November 16, 2020. University of Maryland, College Park.