namma-yatri

name	namma-yatri
description	Build, contribute to, and deploy Namma Yatri — India's open-source, zero-commission ride-hailing platform. Covers Haskell backend, PureScript frontend, Rust services, Beckn/ONDC mobility protocol, Nix-based dev setup, and API integration. Use when working with Namma Yatri, ride-hailing, Beckn protocol, or open mobility.
metadata	{"author":"ankitjh4","category":"transport","tags":"namma-yatri, ride-hailing, open-source, beckn, ondc, mobility, haskell, purescript, rust, juspay"}

Namma Yatri

Open-source, zero-commission, driver-centric ride-hailing platform built by JusPay. Powers 71M+ rides across Bengaluru, Chennai, Kolkata, Delhi, Hyderabad, Kochi, and Mysuru.

• Zero commission — drivers keep the full fare
• No surge pricing — transparent, stable fares
• Open source — AGPL-3.0 licensed
• Beckn/ONDC native — built on India's open mobility protocol

Repository: https://github.com/nammayatri/nammayatri Open Data Dashboard: https://nammayatri.in/open/ Community: https://github.com/orgs/nammayatri/discussions

Architecture

Namma Yatri follows the Beckn Protocol for open, interoperable mobility:

┌─────────────┐    Beckn     ┌──────────────┐    Beckn     ┌─────────────┐
│  Rider App  │◄────────────►│    Beckn     │◄────────────►│ Driver App  │
│    (BAP)    │   search/    │   Gateway    │   on_search/ │    (BPP)    │
│             │   confirm    │              │   on_confirm │             │
└─────────────┘              └──────────────┘              └─────────────┘
       │                                                          │
       ▼                                                          ▼
┌─────────────┐                                          ┌──────────────┐
│   Rider     │                                          │   Provider   │
│  Dashboard  │                                          │  Dashboard   │
└─────────────┘                                          └──────────────┘

Key Beckn concepts:

• BAP (Beckn Application Platform) — consumer/rider side
• BPP (Beckn Provider Platform) — driver/provider side
• Beckn Gateway — routes requests between BAP and BPP
• Registry — service discovery and trust layer

Tech Stack

Layer	Technology	Purpose
Backend	Haskell	Core business logic, APIs, ride matching
Frontend	PureScript	Rider and driver mobile apps
Location tracking	Rust	Real-time GPS tracking service
Notifications	Rust	Push notifications via gRPC + Redis Streams
GTFS server	Rust	Public transit data
Build system	Nix	Reproducible builds and dev environments
Haskell build	Cabal	Package management and compilation
Configuration	Dhall	Type-safe config files
Database	PostgreSQL	Primary data store
Cache	Redis	Session/location caching
Messaging	Kafka	Event streaming between services
Routing	OSRM	Open-source route calculations

Project Structure (Monorepo)

nammayatri/
├── Backend/                         # Haskell backend (main codebase)
│   ├── rider-platform/
│   │   ├── rider-app/               # Rider-facing APIs (port 8013)
│   │   └── public-transport/        # Public transit integration
│   ├── provider-platform/
│   │   └── dynamic-offer-driver-app/
│   │       ├── Main/                # Driver-facing APIs (port 8016)
│   │       └── Allocator/           # Ride matching engine (port 9996)
│   ├── dashboard/
│   │   ├── rider-dashboard/         # Rider ops dashboard
│   │   └── provider-dashboard/      # Provider ops dashboard
│   ├── kafka-consumers/             # Async event processors
│   ├── mocks/                       # Mock third-party services
│   ├── dhall-configs/               # Dhall configuration files
│   └── load-test/                   # Load testing suite
├── Frontend/                        # PureScript frontend
│   ├── ui-customer/                 # Rider mobile app
│   └── ui-driver/                   # Driver mobile app
└── docs/                            # Documentation

Related Repositories

Repository	Language	Purpose
nammayatri/shared-kernel	Haskell	Shared types, utilities, and business logic
nammayatri/location-tracking-service	Rust	Real-time driver location tracking
nammayatri/notification-service	Rust	Push notification delivery
nammayatri/beckn-gateway	Haskell	Beckn protocol routing
nammayatri/common	Nix	Shared Nix configuration

Development Setup

Prerequisites

• Nix package manager (with flakes enabled)
• direnv for automatic shell activation
• ~16GB RAM recommended

Backend Setup

# Clone the repository
git clone https://github.com/nammayatri/nammayatri.git
cd nammayatri

# Activate backend dev environment
ln -sf .envrc.backend .envrc
direnv allow

# Build all backend packages
cd Backend
cabal build all

# Run the full mobility stack (starts all services + DB/Redis/Kafka/OSRM)
, run-mobility-stack-dev

The , prefix is a Nix-provided command runner. Run , alone to see all available commands.

Frontend Setup

# Enter frontend dev shell
nix develop .#frontend

# Rider app
cd Frontend/ui-customer
npm install && npm start

# Driver app
cd Frontend/ui-driver
npm install && npm start

Android builds require Android Studio, google-services.json, and MAPS_API_KEY.

Useful Dev Commands

Command	Purpose
, run-mobility-stack-dev	Start full stack via cabal (recommended)
, run-mobility-stack-nix	Start full stack via Nix build
, ghcid lib/location-updates	Fast compile feedback loop
, run-pgadmin	Start pgAdmin for DB inspection
, run-monitoring	Start Prometheus + Grafana
, kill-svc-ports	Kill all running services
cabal test all	Run test suite

Services & Ports

Service	Port
rider-app	8013
beckn-gateway	8015
dynamic-offer-driver-app	8016
mock-registry	8020
transporter-scheduler	8053
allocation-service	9996

Swagger docs available at http://localhost:<port>/swagger.

API Overview

Rider APIs (BAP — port 8013)

Core rider-side flows following Beckn lifecycle:

Endpoint Pattern	Purpose
/search	Search for available rides
/select	Select a ride option
/init	Initialize booking (address, payment)
/confirm	Confirm and place the booking
/status	Check ride status
/track	Track driver location
/cancel	Cancel a ride
/rating	Rate the driver

Driver APIs (BPP — port 8016)

Driver-side operations:

Endpoint Pattern	Purpose
/on_search	Receive ride requests
/on_select	Respond to ride selection
/on_init	Confirm driver availability
/on_confirm	Accept ride booking
/on_status	Send ride status updates
/on_track	Send location updates
/on_cancel	Handle cancellation

Allocator (port 9996)

The ride matching engine that connects riders with nearby drivers. Handles:

• Proximity-based driver search
• Dynamic offer management (drivers quote their price)
• Ride assignment optimization

Dashboard APIs

Ops dashboards for fleet management, analytics, and administration:

• Rider dashboard — user management, ride history, support
• Provider dashboard — driver management, earnings, compliance

Contributing

Commit Convention

<sub-project>/<type>: #<issue-number> <short summary>

Sub-projects: backend, frontend

Types: feat, fix, chore, ci, docs, perf, refactor, test

Examples:

backend/feat: #341 Driver onboarding flow
frontend/fix: #322 Font size in ride request popup
backend/refactor: #400 Extract payment module

Branch Naming

<sub-project>/<type>/<issue-number><short-description>

Examples:

backend/feat/GH-341/driver-onboarding
frontend/fix/GH-322/font-size

Contribution Flow

1. Find or create an issue on GitHub Issues
2. Fork and create a branch following naming conventions
3. Make changes with atomic commits
4. Ensure cabal build all and cabal test all pass
5. Submit a PR with clear description
6. PRs are typically squash-merged

Code Style

• Haskell: follow existing patterns in the codebase; HLS (Haskell Language Server) configured via Nix
• PureScript: standard PS conventions
• Use direnv + VSCode with recommended extensions for best IDE experience

Key Concepts

Dynamic Offer Model

Unlike fixed-price platforms, Namma Yatri uses a dynamic offer model:

1. Rider searches for a ride
2. Nearby drivers receive the request
3. Each driver quotes their own price
4. Rider selects from available offers

services-flake

External services (PostgreSQL, Redis, Kafka, OSRM) are managed via services-flake, a Nix-based alternative to docker-compose. Services start automatically with , run-mobility-stack-dev.

Shared Kernel

Common types, utilities, and business logic live in the shared-kernel repo. To develop against a local checkout:

# In flake.nix, change:
inputs.shared-kernel.url = "path:/path/to/local/shared-kernel";

Then run nix flake lock --update-input shared-kernel.

Troubleshooting

Issue	Solution
Segfault during linking	ulimit -s 9999 (stack size limit)
Services won't stop after Ctrl-C	, kill-svc-ports
VSCode direnv error	unset NIX_STORE && direnv reload
Slow nix build	Check Nix binary cache is configured
Slow cabal build	Increase jobs in cabal.project or uncomment dev flags
Find Nix dependency source	nix run github:nix-community/nix-melt

References

• Website: https://nammayatri.in
• GitHub Org: https://github.com/nammayatri
• Open Data: https://nammayatri.in/open/
• Beckn Protocol: https://developers.becknprotocol.io/
• ONDC: https://ondc.org
• Founder's Vision (video): https://youtube.com/watch?v=NnyoxiiZLZg
• JusPay: https://juspay.in
• services-flake: https://github.com/juspay/services-flake
• Contributing Guide: https://github.com/nammayatri/nammayatri/blob/main/docs/CONTRIBUTING.md

---

Jai Hind! 🇮🇳