Sep 2025 – Present · 8 mos · India · Remote

• Built developer environments with web based vscode IDE and created a proxy service allowing websocket communications. Improved the latency by 4x by using multi region deployments.
• Introduced redis-based caching to reduce the load on AWS APIs reducing the rate limiting by 15x.
• saved $312000 for the company annually by creating template based instance right sizing and tuning the auto scaling policies

Oct 2023 – Aug 2025 · 1 yr 10 mos · India · Remote

• Developed the v2 APIs for the team that works on a SQL based graph model.
• Developed the consumption logic for the system using AWS SQS
• Developed a weight assignment algorithm, to assign weights to the edges in the asset graph in an efficient way which does not require frequent readjustments when onboarding new edges into the system.
• Improved the Bulk API performance by ~85x from 234s to 3.32s by introducing new indices and using recursive CTE instead of in memory BFS to fetch transitive associations.
• Upgraded the RDS from gp2 to gp3 as a higher IOPS was required which helped us in reducing alarms in the system
• Introduced deletion flow for the ephemeral assetsand vacuumed the database to achieve better performance:
• 57% free disk space
• 3.5x faster queries due to reduced index size
• db cpu utilisation peak reduced to 20% from 60%

Apr 2022 – Sep 2023 · 1 yr 5 mos · India · Remote

• Developed the Asset Attribution system from scratch, the tool helps to identify and attribute the owners/teams to the assets based on multiple relationship betweens assets.
• Developed the APIs for the internal customers to query above data.
• Developed a peak identification system that runs daily after transformations to detect anomalous change in size of data with appropriate thresholds (to reduce false alarms) which had ~90% precision and ~100% recall
• Onboarded multiple new assets into the system increasing the coverage by 25%.
• Improved a complex query from 88 lines of code to ~35 loc which also optimised the transformation runtime from 17 minutes to 7.3 minutes.
• Mentored and onboarded 2 interns and 4 new joinees