5+ years building production backend systems with real operational complexity.
At Spreadd: Apollo Federation v2 with 6 subgraphs, Kafka async pipelines, 87-op GraphQL allowlist (zero-trust), 3 payment integrations, multi-tenant white-labeling, and a Gemini Live API + WebRTC real-time voice bridge (sub-500ms first-audio latency). Cut AWS costs from $480→$240/month and deployment time from 12min→3min.
Freelance: Temporal.io ETL orchestration for multi-tenant SaaS — 11 workflows, 9 schedules per tenant, continueAsNew for unbounded history ETL, defineQuery for live in-flight introspection without polling.
Open to remote roles in distributed systems, backend architecture, and AI integration.
- Temporal.io — workflow orchestration, continueAsNew, defineQuery, per-tenant scheduling
- Apollo Federation v2 — multi-subgraph GraphQL gateway, graphql-shield, Relay compiler
- Node.js, TypeScript — Express, REST, GraphQL
- PostgreSQL, MongoDB — pgvector semantic search, time-series collections, row-level locking
- Kafka, SQS, Socket.io — async decoupling, event-driven pipelines, real-time pub/sub
- React, Next.js 15, React Native (Expo SDK 53)
- React Three Fiber (3D), Zustand, React Router v7, Shopify Embedded App
- Gemini Live API, OpenAI — real-time voice, 3072-dim embeddings, streaming TTS
- WebRTC — PCM audio bridging, FSM session lifecycle management
- Stripe, Razorpay, Web3
- AWS — Lambda, ECS, EC2, RDS, S3, SQS, SNS, SES, Secrets Manager
- Docker multistage builds, CircleCI, GitHub Actions, Jenkins
- 50% AWS cost reduction ($480→$240/month) · 75% deploy time reduction (12min→3min)
B2B influencer marketing SaaS — brands, agencies, creators across Instagram, YouTube, TikTok, Twitter.
- Led team of 8–10 engineers (3 concurrent active workstreams)
- Apollo Federation v2: 6 independent subgraphs, 87-op allowlist, graphql-shield role composition
- Kafka async analytics ingestion; MongoDB distributed rate-limit ledger across container replicas
- 13+ time-series analytics collections; per-user Socket.io rooms; 4 notification channels
- Shared npm library at v5.2.0; 9 background cron jobs; RSA replay-attack prevention
- AI edtech platform (React Native + Gemini Live API + WebRTC): 10ms PCM frame pacer, 45s ring buffer, sub-500ms first-audio latency, FSM for streaming session lifecycle
- Next.js e-commerce platforms with payment gateway, cart, and REST API integrations
- Page load optimization and conversion rate improvements
Multi-tenant analytics ETL with Temporal.io workflow orchestration.
continueAsNewfor unbounded historical ETL (tail-call optimization for workflow history limits)defineQueryfor live in-flight workflow introspection — no polling, no external state store- 11 workflows · per-tenant isolated scheduling · dual-layer graceful timeout
- Full test suite with
TestWorkflowEnvironment· TypeScript · PostgreSQL
Shift-end pick-performance reporting for warehouse sortation robots.
- An AWS Lambda runs 3× per day (at the end of each 8-hour shift)
- Queries MongoDB for robot telemetry
- Generates an Excel report
- Adds an AI-written shift summary via Gemini
- Archives to S3, and sends via SES email.
- Full test suite with · TypeScript · MongoDB
B.E. in Information Technology — Gujarat Technological University (2017)
Advanced Certification in Persuasive UX Strategy — IIT Delhi (2024)
Open to remote engineering roles paying in USD/international rates — distributed systems, workflow orchestration, backend architecture, AI integration.
