Dec 2024 – Present · 1 yr 3 mos · Hyderabad, Telangana, India

• Tech lead of a 3-member squad in the backend re-architecture project of D365 Customer Voice to make the product more compliant and bring it at par with other D365 apps. In parallel, I lead the service monitoring and site reliability pod consisting of 4 engineers.
• Designed a new entity to help our customers understand Sync errors that arise due to permission issues, which is expected to reduce support requests by at least 20/month. Shipped another entity to store the aggregated metric data of responses using a moving average method to prevent time consuming SQL queries.
• Worked with the org’s infrastructure team to onboard to a new shared Service Bus offering. Proposed using another shared Cosmos DB offering as our transient caching service. Both design decisions reduced our product’s projected global infrastructure cost by more than 600k/year.
• Under my leadership, our service monitors detected every outage within the first 15 minutes which contributed to zero customer reported SEV 2s in 6 months. We also reduced noisy monthly SEV 2s to zero from 4 by only triggering high severity alerts when at least a certain percentage of tenants experience SLA drops.
• Initiated DevEx improvements like using local Service Bus Emulator to test pub-sub locally. Introduced load testing using Locust and Python which helped to identify latency issues in our heavy traffic (500 peak RPS) APIs.
• Heavily invested in the overall growth of team members by mentoring interns and new hires.

Sep 2023 – Nov 2024 · 1 yr 2 mos · Hyderabad, Telangana, India

• Contributing to the overall architecture and feature development in both D365 Sales and Customer Voice.
• Immersive Copilot: A full-page copilot experience where real-time and precomputed out-of-the-box insights are visible to the users along with copilot chat. Designed and developed the generic code for hosting any insight and implemented the customer updates insight which showed a quick catchup of relevant sales records. This feature now reports an impressive 20k MAU though still only in preview.
• Prepare for a Sales Meeting: Challenge was to find and show insights from CRM for Outlook meetings of sellers with their customers. Served as the tech lead of a 3-member team from India in this cross-geo collaboration. Responsible for the design and backend implementation where the feature was segmented into parts for which LLM APIs could be called in parallel to reduce latency. Post rollout, MAU reached 7k which was a 30% growth.
• Azure Infrastructure Upgrades: To meet new security guidelines, upgraded Azure services being used in D365 Customer Voice to the latest SDKs which included vast code changes due to new SDK contracts and enforced credential-free authentication amongst all Azure services.
• Mentored new hires, oversaw service deployments and provided technical guidance to the team

Mar 2022 – Aug 2023 · 1 yr 5 mos · Hyderabad, Telangana, India

• Part of the Dynamics 365 Customer Voice product engineering team which is a feedback platform for enterprises.
• Dataverse Sync System: Assumed full ownership of this service which serves 50 million requests monthly after solving long pending bugs and race conditions within the first 2 months which in turn improved SLA from 99.5 to 99.9. Reduced incidents per month from 20 to 5 in just one year.
• Infra Cost Reduction: Decreased function app infra costs by fine tuning the auto-scale rules and did thorough stale resources cleanup which led to $13,000 per month infrastructure cost savings.
• .NET 6 Migration: Spearheaded the migration of services to .NET 6, cleared tech debt wherever possible, reviewed all C# backend migration related PRs, and coordinated testing with the vendor team which made the product compliant with the Microsoft security standards.
• Survey Email Sending Service: During an on-call stint, emails were failing due to 100% CPU usage. Root-caused and shipped the code within a week without any prior knowledge of the service post which became a co-owner.
• Service Monitoring: Due to no service monitors, multi-region outages continued happening. Took charge of this and created azure alerts for all critical use-cases. Documented this thoroughly and organized a team-wide session. This led to zero multi-region outages in the next year and set 15 minutes as the standard incident detection time.