10·Product
Content Factory
A project-management and AI content-production platform, built solo for BARNES Dubai, the Dubai real-estate brokerage where I worked: a three-service monorepo running the full arc of a property content shoot, intake through AI scoring, task assignment, QA and client review, alongside a 27-tool internal catalog for 3D capture, video, AI content and production paperwork.
ResultReal people at BARNES Dubai used this to run photo and video production for property listings end to end: intake, AI coverage and complexity scoring, task assignment across the team, QA, and client review.
[ Ask about this build ]01Develop
What it took
Skills behind it.
Primary discipline plus the support stack.
Skills demonstrated
- Product StrategyPrimary
Designed the seven-stage project pipeline and nine-stage task lifecycle, the RBAC model, and the surface itself: a dashboard, a projects module with per-project tabs for assets, rename plans, prompt packs, tasks and review, a kanban task board, and a 12-tab admin panel.
- Backend EngineeringSecondary
FastAPI + SQLAlchemy 2.0 API (12 Alembic migrations, org-scoped multi-tenant queries, row-locked state transitions) and Celery workers across intake, video and assignment queues, on Postgres + pgvector, Redis and MinIO.
- AI & Machine LearningSecondary
A pluggable AI provider gateway (mock, OpenAI, Anthropic, Ollama, three_d_scene) defaulting to mock so the platform runs without GPU inference, behind coverage/complexity scoring, rename suggestions and creative-brief generation.
- Frontend EngineeringSecondary
Next.js 14 + Tailwind + shadcn/ui web app, including the 27-tool Creator Tools catalog spanning 3D & Spatial, Video & Motion, AI Content, Data & Intelligence and Production.
Develop · Notes
The build, in full.
The decisions on the record - kept off this page so the case study stays a read, not a scroll.
Read the full build notes02Deliver
What shipped.
Real people at BARNES Dubai used this to run photo and video production for property listings end to end: intake, AI coverage and complexity scoring, task assignment across the team, QA, and client review. I no longer work there, and there are no surviving usage records or analytics to cite a number against, so this stays qualitative rather than a stat. It isn't deployed anywhere today. The screens on this page are a deliberate reconstruction made for this case study: the unmodified frontend code, run locally against a small mock API server, loaded from the project's own seed script with names genericized to role labels rather than a capture of the original usage.
By the numbers
Services
Next.js web app, FastAPI API, Celery worker
Creator Tools
3D, video, AI content, data intelligence, production
RBAC roles
client, team, admin, super_admin
Admin tabs
Frames
Selected from the work.
7 frames
01The sign-in screen. 02The dashboard, shown here with representative project data, not a real client's. 03The projects list, spanning the pipeline's stages. 04A project's Overview tab: AI coverage/complexity scores and the team assigned, by role. 05The task board across the full state machine, from Assigned to Done. 06The Creator Tools catalog: 3D & Spatial, Video & Motion, AI Content, Data & Intelligence and Production. 07The Admin panel's Team Members tab, one of twelve; every email is a role label at a demo domain, not a real address.
See also
AI / ML
Barnes Dubai LLM
A domain-tuned language model for a Dubai brokerage, answering in seconds the market-comparison brief that used to take an analyst an afternoon.
Product
Barnes Dubai Dev Portal
The internal cockpit BARNES Dubai's innovation programme runs on: project tracking, analytics and a marketing-engineering toolkit in one place.
AI / ML
Data Centre
BARNES Dubai's internal data tooling: an entity-centric valuation platform and AVM, absorbed into a broker tool over ~840,000 ownership records.