Mar 2025 – Mar 2026 · 1 yr · Bengaluru · On-site

• Architected a On-call AI Agent platform in Python using Google ADK, orchestrating 10+ specialized agents (logs, traces, DB inspection, code & commit analysis) to autonomously investigate production alerts and generate RCA reports, handling 100 incidents/day and reducing manual on-call investigation effort by 60%.
• Built an async FastAPI endpoint with Server-Sent Events(SSE) to stream real-time incident investigation progress, enabling Slack-based human-in-the-loop controls & live agent actitvity updates for long-running RCA workflows.
• Integrated Arize Phoenix for LLM observability and distributed tracing of agent response generation and latency, enabling diagnosis of slow or failed responses, decreasing average investigation time for LLM errors by 50%.
• Optimized database enrichment and entity-resolution pipelines using dependency-aware batch fetching, reducing DB State API calls from 10+ sequential requests to 3–5 batched operations, lowering investigation latency by 40% and reducing backend load during incident spikes.
• Designed cost-aware iterative RCA orchestration, dynamically escalating from low-cost log & DB queries to high-cost code & commit analysis, with stall-detection safeguards cutting unnecessary LLM/tool invocations by 40%.
• Architected and implemented backend integration of Zomato’s Nugget SDK into the Fi Money app, enabling contextual, AI-powered in-app support and enhancing user engagement.
• Designed a streaming RPC-based analytics event orchestrator to stream user interaction events, enabling live monitoring, debugging, and data-driven insights.

Sep 2024 – Mar 2025 · 6 mos · Bengaluru · On-site

• Developed a high-performance consumer RPC that subscribes to an SQS queue and utilises concurrent go-routines for efficient processing, reducing latency by 500ms.
• Optimised stock catalog refresh workflow by designing a Golang script for a scheduled Jenkins job, leveraging keyset pagination (validated via PgSQL query plan) to update 20,000 stocks efficiently while minimising memory usage.
• Implemented Redis write-around caching for DAO operations, reducing database load from 500 QPS to 200 QPS, significantly improving query performance.
• Integrated SDUI for mutual fund and EPF analytics, enabling batch retrieval of multiple analyzer reports in a single gRPC call, cutting response latency by 20%.

May 2024 – Sep 2024 · 4 mos · Remote

• Developed an innovative SaaS solution to automate marketing and sales campaigns, cutting manual workload and reducing operational costs by 30%.
• Engineered a custom role-based authorisation system with NestJS and TypeScript, boosting security and preventing unauthorised access for over 500 daily users.
• Integrated BullMQ for scheduled, high-volume email dispatches and optimized MongoDB aggregation pipelines, enhancing system efficiency by 40% and improving data handling.

Feb 2024 – Aug 2025 · 1 yr 6 mos · Remote

May 2023 – Jul 2023 · 2 mos · New Delhi, Delhi, India · On-site