Hello, I am Anirudh Goel, a former Software Engineer at CERN, and I'm currently working on something exciting that's still under the wraps! I designed & developed this web page and it lists brief highlights of my professional journey. Apart from technology, I like to talk about finance (investing, fintech), business, education and baking.
Internships & Experiences
Core member of the Database on Demand (DBoD) Team in the CERN IT department; DBoD runs 1000+ database instances of PostgreSQL, MySQL and InfluxDB.
- High availability for PostgreSQL based critical services
- Setup automated pipeline (using RunDeck, Docker, Nomad & Puppet) used to deploy multiple highly available PostgreSQL clusters for critical services
- Wrote custom bash scripts for failover and online recovery. Patch also accepted for bug fix in official PgPool failover script.
- Used synchronous replication to guarantee zero data loss; PgPool-II for load balancing and failover management; deployed using Nomad to ensure redundancy on PgPool level; used PgPool-II exporter, Prometheus and Prometheus alertmanager for real-time monitoring and alerts.
- Active member of the PgPool hackers community; discovered and contributed to patching multiple bugs in the PgPool-II project.
- Secondary Level Backup & Recovery System [Presented in HEPiX Autumn 2021]
- Deployed using Puppet, a Python based service RPM that compresses (using zstd), encrypts and uploads new snapshots of 730+ database instances to EOS (CERN network storage) for ensuring a secondary layer of backups; 7.5 TBs of highly compressed (~80%) data backed up everyday.
- Developed an automated restore service to continuously restore and test backups, for validating backup & restore pipelines for a reliable Point-in- Time Recovery in case of issues. Service helped uncover various issues - configuration mis-match in config file vs runtime, permission issues in database cluster, WAL level errors, etc.
- Developed AV Analytics
- Django based platform that records the complete journey of all users across company’s multiple platforms.
- This data has helped to increase the customer acquisition by 54% and customer retention by 40%.
- Accrued 85 Million records (2 Billion data points) till date, that are being used for various purposes including analytics and behavioural analysis.
- The service has been scaled from few hundred to few thousand users per day without requiring any changes.
- Setup pipeline using AWS Kinesis Data Stream to sync the users’ profiles across platforms to obtain a 360 degree view of all users.
Open Source Projects
PinAutomatic
This is a webapp to pin all images of one Pinterest Board to another automatically. Hosted on Heroku, it uses Heroku Postgres as Database, flask-user package for User Management, Redis To Go server to store the pinning tasks in a queue.
Started as a hobby project, it has been approved by Pinterest is currently being used by 100+ users and has re-pinned 200,000+ pins till date.
