Umakant Vashishtha

Apple Inc.

Hyderabad, India

• Built a self-serve platform for creating, testing, and deploying AI agents without engineering dependency — featuring native integration with Apple's internal data providers, on-the-fly MCP server creation/updates, and a no-code interface for business users to own the full agent lifecycle including version management & CI/CD.
• Built "Autopilot," an Electron desktop app that aggregates action items across Slack, task managers, calendar, and meetings into a unified view. Integrated Claude as an on-device task execution layer with fully local memory persistence, enabling users to delegate coding tasks, writing PRs and testing apps to Claude Code.
• Designed an AI agent that interprets natural language queries against live Tableau dashboards and generates DAG workflows for async report automation. Built the underlying Task Orchestration engine to schedule, execute, and monitor DAG pipelines.
• Contributed to an internal Enterprise Knowledge Graph platform, including a purpose-built graph UI for exploring entity relationships and an MCP server layer for LLM-powered graph summarization.

Razorpay

Bengaluru, India

Worked as Backend Engineer at RazorpayX

• Led architecture and development of a vendor KYC microservice with approval workflows using Cadence & Cassandra DB, handling 500+ RPS for 1500+ vendors across 10+ merchants. Set up on-call processes with metrics and alerts with Prometheus/Grafana dashboards/PagerDuty.
• Created Razorpay's VS Code Extension for AI-powered payment integration — LinkedIn, YouTube — enhancing development experience for clients to easily migrate their existing integration to Razorpay.
• Developed pricing/billing systems with atomicity and retry mechanisms, improving fee recovery efficiency by ~80%. Built a performance dashboard using Looker.

• Led big-bet projects like multi-Current Accounts for 30+ enterprises and migrating ~50k customers to a new Pay-On-Behalf-Of payout scheme under RBI guidelines.
• Improved slow APIs by optimizing SQL queries and reducing latency from 8s to 200ms using observability with Go's pprof, metrics and traces.
• Designed and delivered a collaborative platform for Current Account onboarding, boosting conversion by 45% through automation and internal tools.
• Migrated our team's code from PHP monolith to Golang microservice, supporting 70k+ merchants.

Technologies Used:
Golang, Kubernetes, Docker, MySQL, Cassandra, Cadence, PHP (Laravel), Apache Kafka, Grafana, Prometheus, PagerDuty, SumoLogic, HyperTrace, React, Node.js, Salesforce, Looker

Butterpaper / Go4Life

Gurgaon, India

• Built a full agri-tech e-commerce SaaS platform from 0 to 1 — architecting a BFF layer on GraphQL/Node.js + Postgres serving a React Native customer app, React admin dashboard, React Native delivery app, and Metabase analytics dashboard across mobile and web clients.
• Scaled from solo founder-engineer to leading a team of 4 (2 engineers + 2 interns) — owning the full product lifecycle: design, development, deployment, and monitoring. The platform grew to 20k+ active customers across ~10 clients at peak throughput of 2,000+ RPM.
• Built a real-time delivery management system with geo-based order assignment, live rider location tracking, and an ops dashboard for monitoring deliveries in flight.
• Re-architected the platform into a multi-tenant SaaS model, enabling custom-branded app builds per organization that could go live on the Play Store and App Store within days.

Technologies Used:
Node.js, TypeScript, PostgreSQL, React, GraphQL (Apollo), React Native, MongoDB, Metabase, AWS (SES, EC2, Lambda, Load Balancer), Nginx, Prisma, Sequelize

CredFlow

Delhi, India

• Owned end-to-end design and implementation of 8+ core product features in a 3-person engineering team — including Subscription Management, CRM, Analytical Dashboards (Sales, Collections, ACP trends), Incentive Scheme Management, Payment Projection, and an SMS Credit System — serving 100+ organizations processing 500k+ invoices and 100k+ tasks every day.
• Proactively identified and patched security vulnerabilities, preventing active threats from a malicious actor and hardening account protection across the platform.
• Automated email forwarding workflows using AWS SES and Lambda, reducing manual ops overhead.

Technologies Used:
Node.js, PostgreSQL, React, Electron, Message Queue, Puppeteer, AWS (SES, EC2, Lambda)

Masai School