Google Core ML Infrastructure - Optimizations for experiments & workloads across Google’s Accelerators - add features for tools serving Gemini, Deepmind, Waymo, Ads, YouTube, Search and smaller google PAs

YouTube Core Reliability Slow optional dependenciesShorts Reliability - Track and fix watch session failures for Android Shorts- added 17 client error telemetries with context data - analyzed and triaged the most impacting cases- resolved 2 most impacting cases which led to X% improvement in global Shorts Watch TimeClient Application Error logging Improvements - identified several issues with error logging across YouTube’s client apps- initiated a… Show more YouTube Core Reliability Slow optional dependenciesShorts Reliability - Track and fix watch session failures for Android Shorts- added 17 client error telemetries with context data - analyzed and triaged the most impacting cases- resolved 2 most impacting cases which led to X% improvement in global Shorts Watch TimeClient Application Error logging Improvements - identified several issues with error logging across YouTube’s client apps- initiated a YT-wide effort to standardize logging metadata for consistent uniform data (approved by 16 teams)- improved triage time by attributing errors to screen types & user experiences- reclassified the severity of error cases which improved metric noise by 75%- added a real-time error monitoring signal for YT Music & YTTV that caught 5 failure regressions over 4 months- configured pre production jobs to use error signals to detect and mitigate client side regressions before they get to production Stuck RPC Metric - Monitoring metric that tracks the number of unary and streaming RPCs stuck in a Loop of Death- Real-time graphs that show each service’s Stuck RPC status- Setup Alerting to notify OnCallers when persisting queries enter a loop of death - Documented a handy Playbook to triage and mitigate issues where a service experiences high number stuck RPCsMonitoring Consoles Migration- Migrated monitoring consoles from an old Internal tool to a new Internal tool Load Balancer CPU Utilization- Saved ~2 SWE/yr of resources by testing and increasing CPU utilization of YouTube’s load balancing Frontend layersDegradation monitoring of Optional DependenciesProblem: Optional dependencies of services are soft dependencies that don’t necessarily fail or break a service. However, these dependencies can serve successful degraded responses that can have UI visible and revenue impactSolution: Integrated Degradation Monitoring for user and revenue impacting features on YouTube’s FrontEnd Service. Show less

Worked on a Service that connects Medial Donors to Patients in need of monetary assistance for their medical bills. We do this by providing donors accurate Public Electronic Health Records (EHRs) from Hospitals available through EPIC Systems

- Fostered an inclusive community among 2200 residents, maximizing positive resident interactions- Mediated and provided conflict resolution in a professional and timely manner

Research Assistant under Professor Udaya Shankar. I defined a service that builds on the popular 2 Phase commit (2PC) Database Concurrency Control Protocol. The protocol processes database transactions reliably and atomically in a distributed environment. Dealt with a single point of failure scenario where the coordinator node crashes. I started with a stub implementation followed by a service tester for testing applications and implementations.

Programmed, designed and deployed an Agile Web App that enabled students to interact with lab equipment remotely. The project was initiated in response to the COVID-19 pandemic.The front of the Web Application was developed using the React framework and the backend was developed using Node.js. The wireframe and UI/UX of the Web App were designed using Figma. The user database and sessions were maintained using MongoDB. Finally, the Web Application was deployed on Heroku.The Web… Show more Programmed, designed and deployed an Agile Web App that enabled students to interact with lab equipment remotely. The project was initiated in response to the COVID-19 pandemic.The front of the Web Application was developed using the React framework and the backend was developed using Node.js. The wireframe and UI/UX of the Web App were designed using Figma. The user database and sessions were maintained using MongoDB. Finally, the Web Application was deployed on Heroku.The Web App is defunct as labs are no longer online Show less

Taught Mathematics, Physics, English, and Programming subjects to students online and in-person. While most of my students I tutored were College Students, some of them were High School students that needed help with Standardized Tests like SAT and IELTS. I also tutored College students on Algorithms and Object Oriented Programming.

Data analysis and visualization of employment rate, literacy rate and dropout rate for each state.Core data provided by National Commission on Statistics.In response to the findings, I also led presentations focused on familiarizing High School dropouts about employment opportunities in the Electronics Sector of India.I was tasked with reaching out to new schools to expand the organizations’ reach to High or Middle School Dropouts. Lastly, I initiated and worked on the idea of… Show more Data analysis and visualization of employment rate, literacy rate and dropout rate for each state.Core data provided by National Commission on Statistics.In response to the findings, I also led presentations focused on familiarizing High School dropouts about employment opportunities in the Electronics Sector of India.I was tasked with reaching out to new schools to expand the organizations’ reach to High or Middle School Dropouts. Lastly, I initiated and worked on the idea of establishing an online platform for educating aspiring candidates. Show less