Jalandhar, Punjab, India
🛠️ Backend & full-stack engineer with 10+ years building failure-prone distributed systems where correctness and latency both matter. 🚀 WHAT I'VE SHIPPED IN PRODUCTION 📍 Two-tier geospatial alerting (Redis GEO + PostGIS) — sub-500ms proximity lookups over 100k+ location events/day, 10k+ concurrent responders (Intterra, emergency management SaaS). 💸 Payroll automation syncing 10k+ employees across 300+ companies through Kafka-driven pipelines with idempotent, retry-safe handlers (Dapt Tech). 🐳 Sandboxed code execution in Kubernetes-orchestrated Docker containers — 20–100 concurrent runs, hardened against filesystem/network escape (Mastery Coding). 🚗 Expanding-radius driver matching with live ETA via Socket.io — 200–1,000 concurrent ride requests (Qudos). 🤖 LLM-augmented workflows with LangChain + LangSmith evaluation gating low-confidence outputs before they reach end users. ⚙️ HOW I OPERATE Own backend architecture end-to-end — API contracts, data models, event schemas, deployment, observability. Lead 5-10 engineer backend teams. Ship frontend independently when the product needs it (Next.js / React portals across four shipped products). 🔧 FAILURE-DRIVEN DECISIONS • Dapt Tech's synchronous sync dropped events silently under load → re-architected to Kafka with idempotent handlers. • Intterra's single-tier spatial queries degraded at scale → introduced Redis GEO as hot layer with PostGIS fallback. 🌐 DOMAINS Emergency management · Fintech/payroll · Edtech · E-commerce · Mobility · Sports analytics. 🧰 STACK Node.js · Nest.js · Fastify · Express · Python (FastAPI/Django) · PostgreSQL + PostGIS · MongoDB · Redis · Kafka · Kubernetes · AWS (Lambda, Step Functions, SQS) · Azure · LangChain · LangGraph · LangSmith · Next.js · React.
Lead backend engineer across multiple production SaaS products. 🔹 Intterra (emergency management SaaS): Architected two-tier geospatial alerting — Redis GEO hot layer + PostGIS for complex spatial queries. Sub-500ms lookups over 100k+ GPS events/day, 10k+ concurrent users. LLM-powered incident summarization via LangChain, with LangSmith evaluation gating low-confidence outputs before they reached commanders. 🔹 Dapt Tech (HR/payroll SaaS): Replaced manual invoice/job/time-entry sync with Kafka-driven event pipeline. Per-sync time dropped from 10–15 min manual to <5 min automated. Idempotent handlers prevent duplicate payroll under Kafka replays. Scale: 10k+ employees, 300+ companies, millions of events. 🔹 Mastery Coding (edtech): Sandboxed Python execution in Kubernetes-orchestrated Docker containers with enforced CPU/memory limits, 20–100 concurrent runs. Hardened against filesystem/network/os abuse. LLM feedback gated through confidence evaluation. 🔹 Protax Canada (tax/bookkeeping): Tax processing backend unifying Stripe billing, E-sign flows, and Twilio notifications for Canadian tax compliance. Built client-facing Next.js portal for document upload and status tracking. 🔹 Lucky Rhino (corporate rewards): Order-inventory reconciliation APIs linking Shopify/BigCommerce to reward programs with idempotent webhook handling. Owned React/Next.js retailer portal. 🔹 All The Artists (headless e-commerce): Replaced monolithic Shopify theme with decoupled Next.js storefront + Sanity CMS. Static generation + edge caching for sub-second loads. 🧰 Stack: Node.js, Nest.js, Fastify, PostgreSQL + PostGIS, Redis, Kafka, Kubernetes, LangChain, LangSmith, AWS, Azure, Next.js, React, Python.