SARVESH MISHRA

Architected and led the migration from AWS AI Gateway to open-source MLflow; implemented multi-provider traffic routing and token-usage governance, which reduced annual LLM costs by 30%.

Developed an agentic AI layer within Jenkins to autonomously execute multi-repository operations, including PR creation and dependency updates, decreasing developer overhead by 20%.

Designed and built an in-house React + GoLang OnCall application, replicating OpsGenie functionalities for scheduling and notifications, resulting in annual savings of $24K.

Developed and deployed a GPT-powered internal chatbot, enhancing customer engagement by 12% and reducing support ticket resolution time by 5%.

Optimized Kubernetes cluster costs by integrating KRR and OpenCost, right-sizing CPU/memory workloads to reduce infrastructure costs by 38% while maintaining 99.99% uptime.

Spearheaded the migration of compute workloads to ARM instances on AWS, cutting monthly infrastructure spend by 30% and boosting backend service performance by 20%.

Developed reusable Terraform modules for AWS services, standardizing infrastructure-as-code practices and reducing deployment time by 25%.

Implemented Spotify Backstage as a centralized developer portal, improving developer productivity by 25% and decreasing onboarding time by 40%.

Built comprehensive monitoring, logging, and alerting systems with Grafana Stack and distributed tracing, reducing incident response time by 30%.

Re-architected the React SSR service into a Dockerized Node.js renderer, decoupling it from the C# backend and reducing cold-start latency by 42%.

Implemented Redis-integrated caching for SSR React components during high-traffic builds, improving Time To First Byte (TTFB) by 35% and reducing backend CPU usage by 20%.

Integrated Grafana Faro for real-time client-side telemetry, capturing Web Vitals and JS errors, which improved frontend issue resolution by 35% and enabled end-to-end tracing.

Led the migration from legacy RabbitMQ to Kafka, developing an internal messaging library that improved data delivery reliability by 15% and reduced consumer lag by 22%.

Designed and implemented a Redis-backed distributed rate limiter, increasing API revenue by 18% and improving API health score by 25% by throttling abusive traffic.

Built a full-stack React + .NET drag-and-drop website builder for car dealers, achieving 20% faster deployment times and real-time CarWale data integration.