Description

July 2024 - July 2025

  • Achieved a 90% reduction in homepage load time (internally: Munshi -> MunshiV2) by redesigning the Java backend and migrating the internal Cassandra impressions database to a new schema — instead of an individual lookup per view, a single table now holds all views, replacing per-view lookups with a unified table storing total views per user. This cut load time from 40 seconds to 2-3 seconds; verified using tracevault via the request trace id.
  • Introduced org awareness (internally: “Munshi Org Aware”) to the trending table of the Cassandra database by creating a new table with altered schema, allowing each isolated org in the instance to have separate trending objects — orgs are independent shells within ThoughtSpot.
  • Executed team-led zero downtime data migration of 50M+ rows across clusters in Cassandra database by implementing a Java background migration service that safely transferred data and seamlessly switched over to the new schema.
  • Redesigned the user onboarding flow (internally: “Edge Onboarding”) to highlight AI-powered Spotter, guiding users through creating connections, importing data, and asking questions via natural language — introducing Spotter and TSA to a first-time free-trial ThoughtSpot user, aiming to reduce user dropout, resulting in improved feature adoption by 30%.
  • Addressed a security vulnerability in the share API by implementing a Redis-based rate limiter after penetration testing exposed a missing rate limit, effectively controlling API request rates and mitigating abuse risks.
  • Diagnosed and patched a Mixpanel overtracking issue ahead of a renewal, reducing test event volume by 98.5% (20M to 300K) and saving $30K — that figure is what might have gone toward the new Mixpanel plan otherwise (final purchase was $23K, versus a $50K range without the fix). Also fixed several Mixpanel bugs firing at the wrong time or showing the wrong display name.
  • Fixed an Apple security issue by removing cronJobExpression from session info.
  • Reduced Mixpanel Monthly Tracked Users (MTUs) in the testing project from 700k to 10k weekly users by identifying cases where Mixpanel should not be enabled.
  • Added frontend support for monitor alerts in Slack, enabling users to receive data updates directly in Slack without opening the BI tool, increasing notification click rate by 20%.
  • Worked on improving the product homepage, and wrote Jest and Playwright tests to increase coverage.