( Case 03 / 10 ) — India's Biggest IT Market, Indexed

Nehru Place

The studio's largest build to date — a ground-up directory and trade platform for Delhi's Nehru Place IT market. 70+ routes over a 40+ table schema: towers modelled floor by floor, an RFQ engine, encrypted buyer–seller chat, a concierge assistant and a full seller dashboard.

RoleFull-Stack Development
StackNext.js · Three.js · Drizzle ORM · Postgres · Neon · Auth.js
YearAugust 2026
Visit live site↗
( The build )

Nehru Place had no live site with the right theme and none of the structure a market this size needs. We kept the visual identity — the brass-and-ink palette, the mono-type data language — and rebuilt everything underneath: a real schema for towers, floors and a 200+ node category taxonomy across 400+ brands, seller signup with a verification ladder, and a moderation queue. Then we went well past the directory. Buyers post an RFQ, sellers quote against it, and the two talk over end-to-end encrypted chat. Sellers get their own dashboard for offers, leads, reviews, disputes, billing and analytics on a four-tier plan ladder. Staff get catalog de-duplication, product matching, a verification queue and an audit log on every write. A concierge chat dock and a command palette answer visitor questions from one shared corpus — deliberately retrieval-only, with no language model anywhere in the path, so it can never invent a price or misstate a dealer's stock; when it cannot answer, it escalates to a person. The build runs on two separate design registers — an award-tier showcase surface for buyers and search, and a fast, zero-motion product surface for sellers managing listings — so neither slows the other down. Bilingual in English and Hindi, DPDP-compliant by construction, and covered by 40+ test files.

( In the build )
  • 01Tower-by-tower, floor-by-floor directory over a 200+ node taxonomy and 400+ brands — modelled in a real database, not a flat list
  • 02RFQ engine — buyers post, sellers quote, both compare — with end-to-end encrypted chat between them
  • 03A concierge chat dock and command palette sharing one corpus — retrieval-only, no model, with escalation to a human
  • 04Seller dashboard: offers, leads, reviews, disputes, billing and analytics, on a four-tier plan ladder
  • 05Admin console for moderation, verification, catalog de-duplication, product matching and ad campaigns, audit-logged throughout
  • 06A price index built on real price history, saved items, guided tour and site-wide search
  • 07Bilingual (English/Hindi), TOTP two-factor auth, and DPDP compliance built in — consent, data export, erasure and retention jobs
( What changed )

The market is now modelled the way it actually works — building by building, floor by floor — with the full trade loop built and walkable end to end: signup, verification, listings, RFQs, quotes and chat. 70+ routes, 40+ tables and 40+ test files in, dealer onboarding is the next step, not the next build.