Ravi Kumar

Founding Engineer

Bengaluru, Karnataka, India

Zoober

Hyperlocal Commerce Platform | Node.js · MongoDB · AWS

• Built web-based Admin and CRM panels for merchant management, order analytics and real-time monitoring — deployed on AWS (EC2, ALB, CloudFront, S3) achieving 99.9% uptime
• Designed Express.js REST API with optimised MongoDB collections — 2dsphere geospatial indexes for <100ms radius-based store discovery and real-time WebSocket sync for order/rider position updates
• Integrated payment gateways (Juspay, PhonePe, Razorpay) with webhook HMAC-SHA256 verification and idempotent order handling; managed CI/CD pipelines for zero-downtime deployments

MCPVave

AI/GenAI Engine — Model Context Protocol Client across 6 Platforms

• Built MCP client enabling standardised LLM-to-data communication (Claude, GPT, Gemini) via SSE and STDIO; developed mcp_client and llm_shared as internal packages — powering inventory management, demand forecasting and automated customer support

Founding Engineer

Bengaluru, Karnataka, India — Groceries Delivery in minutes

• Architected Node.js/Express backend processing 80K+ orders at <200ms P95 latency; built web-based Admin, CRM and Control Panel for high-level operations and real-time monitoring
• Designed real-time WebSocket sync layer for order/rider state, CRM message exchange and alerts with automatic reconnection; deployed on AWS + GCP with CI/CD pipelines
• Integrated Google Maps, Firebase, ONDC, payment gateways (Juspay, PhonePe, Razorpay, BillDesk); built Dart ERP sync on GCP Cloud Run syncing 35+ stores × 12K+ products

Software Engineer (Remote)

workrush.co.uk

• Developed backend systems in Node.js and PHP; collaborated across time zones on API design, feature specifications and production debugging
• Built consumer-facing mobile apps and shipped 15+ releases across Play Store and App Store

Real-Time Monitoring Dashboard | Event-Driven Microservices Architecture

Python, FastAPI, Next.js 16, React 19, TypeScript, Tailwind CSS v4, Kafka, PostgreSQL, Redis, Docker, OCI (Ampere A1)

• Architected 12-service event-driven system with Kafka-based data pipeline (KRaft mode, Snappy compression) processing live market data from Kite WebSocket API through 4-stage flow (data-ingestion → candle-builder → rsi-calculator → divergence-detector) with <50ms end-to-end latency
• Implemented incremental RSI calculation using Wilder's smoothing (avoiding 300+ candle recalculation per tick) and ZigZag peak/trough analysis for price/RSI divergence detection with Redis pub/sub broadcasting alerts to WebSocket clients and PostgreSQL time-series persistence
• Built multi-source news aggregation (Livemint, Moneycontrol, Financial Express) with keyword tagging (150+ terms), sentiment analysis, and 5-minute Redis caching for real-time financial news delivery
• Developed responsive frontend with Next.js 16 App Router, React 19, TypeScript, Tailwind CSS v4, SWR caching, and embedded TradingView charts for live market visualisation

Agent-Centric Property Listing Platform | Geo Search + AI Chat

NestJS, Next.js, PostgreSQL, PostGIS, Redis, Cloudinary, OCI ARM, Claude AI, Docker, Tailwind CSS, TypeScript

• Built full-stack property listing platform with NestJS + Prisma REST API, PostGIS powered geo-search, and Next.js + Tailwind CSS frontend — connecting property seekers with verified agents in Bangalore
• Implemented complete visit booking lifecycle (pending → scheduled → visited → confirmed/cancelled), agent time-slot management, role-based access (user/agent/admin), and rate-limited SMS OTP auth with daily password fallback
• Integrated Claude AI conversational assistant for natural-language property search; deployed on OCI ARM free tier with Docker + Cloudflare Tunnel — zero infra cost, scalable to 50K+ users

Self-Hostable Push Notification Platform | MCP-Powered for AI Agents

Node.js, TypeScript, Express 5, Prisma, PostgreSQL, BullMQ, Valkey, Flutter, MCP, FCM, SigNoz (OTel), Sentry

• Built a self-hostable, multi-tenant push notification platform with MCP (Model Context Protocol) support — letting AI agents (Claude Desktop etc.) send FCM push notifications directly to users' phones via natural language
• Architected horizontally scalable system with BullMQ on Valkey for async delivery queuing, PgBouncer for connection pooling, and separate stateless API / MCP / Worker processes — each independently scalable
• Shipped full Flutter mobile app for device registration, channel management, and notification history; wired end-to-end observability with SigNoz (OpenTelemetry) + Sentry

Offline-First Mobile App | Flutter + BLoC

• Built cross-platform Flutter app with offline-first architecture using Isolate-based multithreading for background data processing without blocking UI
• Implemented Hydrated BLoC with Hive persistence for offline-first state management, enabling full app functionality without network connectivity

AI Image Generation Pipeline | TypeScript · Node.js · OpenAI · Gemini

• Automated pipeline using OpenAI API for runtime prompt generation and Gemini for high-quality product image generation — output integrated into React/Next.js e-commerce product showcase pages

Flutter Icon Libraries | Static + Animated | lucide-animated Ecosystem

• Publishing and maintaining two Flutter packages: flutter_lucide (1,666+ static icons) and flutter_lucide_animated (smooth Lottie-style animations) with tree-shaking optimization
• flutter_lucide_animated officially featured in pqoqubbw/icons — the lucide-animated ecosystem alongside Svelte, Vue, and Angular implementations

AI Image Generation Pipeline | E-commerce Product Visuals

TypeScript, Node.js, OpenAI API, Google Gemini, Nano Banana

• Built automated pipeline using OpenAI API for runtime prompt generation and Gemini Nano Banana model for high-quality realistic image generation — output integrated into React/Next.js e-commerce product showcase pages
• Designed for e-commerce/website product showcase generation with optimised prompts for professional product photography