Tapri.AI is a social community app exclusively for India's delivery riders and gig workers — Swiggy, Zomato, Zepto, Blinkit. Location-tagged feed, local rider groups, real-time traffic + police-check alerts, and rider-made reels. Built Hindi-first so riders actually use it on the road.
Gig riders earn in silos — WhatsApp groups fragment, platform apps don't share rider intel, and critical road info (checkpoints, demand surges, outages) reaches riders too late.
A mobile-first community layer: location-tagged posts, reels, and rider groups; push alerts for local demand + traffic; direct messaging; platform-agnostic earning strategy sharing.
Tapri.AI is a Hindi-first social network for delivery riders, combining a location-tagged feed, rider groups, reels, direct messaging and push alerts for traffic and demand. A React Native app talks to a modular Node.js backend on AWS, with PostgreSQL as the system of record, Redis for feed fan-out and presence, and Firebase for push notifications; a Next.js site handles the public web and blog.
Riders use a React Native app built Hindi-first for one-handed use on the road; the public web and blog run on Next.js.
An API gateway terminates auth and rate limits, while a WebSocket layer streams feed updates, chat and live alerts.
Node.js services are split by domain so feed ranking, messaging and alerts can scale independently.
PostgreSQL stores users, posts, groups and geotags; Redis backs timelines, presence, counters and pub/sub for real-time delivery.
Reels and photos are uploaded directly to object storage, transcoded asynchronously, and served via CDN; Firebase delivers push alerts.
Runs on AWS with containerised services, background workers and CI/CD, sized for spiky evening delivery peaks.
A rider opens the app; the gateway validates the JWT and the feed service returns a timeline from Redis keyed to the rider's city.
The rider records a reel; the app uploads it straight to object storage using a presigned URL and posts the metadata to the API.
A transcoding worker generates mobile-friendly renditions and thumbnails, then marks the post ready for the feed.
The feed service fans the post out to followers and nearby riders via Redis, and WebSocket clients receive it live.
A rider posts a police-check or traffic alert tagged to a location; the alerts service notifies riders in that radius via Firebase push.
Riders discuss it in a local group or by direct message, with messages persisted in PostgreSQL and delivered over WebSockets.
Riders refresh constantly between orders on weak networks, so precomputed per-city timelines in Redis give instant loads while PostgreSQL remains the durable record.
Routing reels through API servers would saturate them; presigned uploads to object storage with async transcoding keep the API thin and uploads resilient.
Firebase Cloud Messaging solved device tokens and delivery across Android fleets cheaply, while core social data stayed in PostgreSQL for ownership and complex queries.
Hyper-local road intel is the product's edge, so every post and alert carries coordinates indexed in PostgreSQL, enabling radius queries for feeds and notifications.
“CodeLamda shipped Tapri as a production-grade social app — reels, real-time feed, push alerts, in twelve weeks. Hindi-first UX for riders who earn on the road, exactly how we briefed it.”
We ship MVPs, AI agents, and production systems in 8 weeks. Bring the rough spec — we'll scope the build inside a week with a fixed timeline and quote.
