Marketplace API
AutoApp Backend
NestJS car marketplace API with Elasticsearch faceted search, Redis caching, Prisma/PostgreSQL, and dealer/CNIC verification workflows.
Background & Context
Car marketplaces in Pakistan live or die on search speed and trust. Buyers filter by make, model, city, and price at volume; dealers need verification; fraud patterns appear as soon as listings scale. AutoApp required an API foundation that could serve large listing catalogs without collapsing under faceted queries—and encode CNIC/dealer trust into the domain rather than bolting it on later.
Tekvers engineered a NestJS modular monolith (autoapp-backend): auth, users, dealers, listings, search, favorites, reports, admin, and health modules. PostgreSQL via Prisma holds transactional data; Redis caches popular listings; Elasticsearch powers text relevance plus facet aggregations with sync-on-write and a nightly reindex cron. Docker Compose runs API + dependencies for production parity.
Honesty for operators: phone OTP is mock (fixed code logged to console), image fields are URL lists without an upload pipeline, Prisma migrations/ are not checked in yet, and dealer registration does not auto-promote User.role to DEALER. The deliverable is a strong API foundation with clear hardening remaining before consumer-scale production.
The Challenge
A Pakistan car marketplace needed faceted search, dealer and identity verification, favorites/reports, and infrastructure that can grow with listing volume—from an API-first modular backend. Without a coherent marketplace api foundation, AutoApp Backend stakeholders faced fragmented tools, slow handoffs, and limited visibility—classic failure modes Tekvers designs against.
The Solution
AutoApp Backend: NestJS car marketplace API with Elasticsearch faceted search, Redis caching, Prisma/PostgreSQL, and dealer/CNIC verification workflows. Production-shaped API scaffold for high-volume car listings. Tekvers delivered a maintainable marketplace api system for AutoApp Backend using NestJS, PostgreSQL, Redis, Elasticsearch, with phased rollout, operator workflows, and documentation suited to long-term ownership.
Our Approach
Search & cache architecture
Elasticsearch handles faceted listing discovery (cities, makes, models, fuelTypes) while Redis absorbs hot popular-listing reads. Listing create/update/delete syncs ES; admins can reindex; cron reindexes daily at 03:00.
Trust & marketplace modules
CNIC submit with format validation and admin verify; dealer register (business name + city) with admin verification/subscription plan patches; listing reports with admin status workflow; favorites per user.
Auth, RBAC, and ops
Phone OTP request/verify (mock SMS), JWT + hashed refresh token rotation, global throttling and DTO validation, Pakistan phone validation (`@IsPhoneNumber('PK')`), and `GET /api/health` across Postgres, Redis, and Elasticsearch.
What We Delivered
- NestJS modular monolith API under `/api`
- Prisma/PostgreSQL persistence for users, dealers, listings, and related entities
- Elasticsearch faceted search with sync + nightly reindex fallback
- Redis caching for popular ACTIVE listings
- OTP auth (mock) with JWT access/refresh and role guards
- CNIC and dealer verification admin workflows
- Favorites and listing reports modules
- Docker Compose stack for API + Postgres + Redis + Elasticsearch
Outcomes & Impact
- Production-shaped API scaffold for high-volume car listings
- Faceted search foundation suitable for marketplace UX
- Trust workflows (CNIC, dealers, reports) modeled in the core domain
- Clear hardening backlog: real SMS OTP, migrations, image upload, role auto-promotion
AutoApp’s marketplace can search, verify, and scale from a Tekvers API core—facets stay fast, dealers and identity are enforceable by admins, and ops can reproduce the dependency stack locally with Docker Compose.
Technology Stack
Services Delivered
Related Case Studies
Guides & articles from this niche
Planning a similar platform? Share your scope and we will map architecture, delivery phases, and a realistic timeline.