Full-Stack Developer & System Architect
GigsHub
AI-powered freelancer marketplace—Django REST backend, Next.js dashboard, and a FastAPI + Claude microservice that analyses every project and ranks every bid across five weighted dimensions.
Nairobi · remote delivery · live marketplace platform
The problem
Most freelancer platforms treat matching as a search problem—keyword filters and star ratings. That produces noise. A client posting a React dashboard should not sift through dozens of bids to find freelancers who actually understand multi-tenant architecture and WebSockets. Freelancers need fair positioning based on skill depth and track record, not who bids first or lowest. Operators need visibility across projects and payouts without living in spreadsheets.
The client need
A production marketplace that automates the two critical decisions: who should work on this project, and which bid the client should choose—with scored reasoning, not guesswork.
What I built
I designed and built GigsHub from scratch: a Django REST core for marketplace operations, a Next.js admin and user dashboard, and a separately deployed FastAPI AI microservice powered by Claude (Anthropic). The AI engine analyses every project on submission, extracts required skills (including implied ones), scores the freelancer pool, and ranks bids across five weighted dimensions.
Main platform backend (Django 5 + DRF)
- Multi-tenant JWT auth (SimpleJWT) with RBAC for clients, freelancers, and admins
- Polymorphic service models for software, research, and custom project types
- Full project lifecycle—submission, bidding, assignment, workspace, status, order history
- Django Channels WebSocket chat inside project workspaces
- Payout management, document handling, support ticketing, and resources as separate apps
- Swagger-documented REST API with API-key gateway
AI intelligence engine (FastAPI + Claude)
- Automatic project analysis on submission—skills extraction, complexity, experience level, ranked freelancer shortlist with reasoning
- Bid ranking across five dimensions: skill match 40%, proposal quality 20%, experience 20%, ratings 10%, price 10%
- Cached scores so admin bid review loads instantly
- Signals-based freelancer pool sync—bulk bootstrap plus incremental updates on every profile save
Tiered freelancer marketplace
- Native — Partner-Selective: rare niche skills, scarcity-driven top-of-market positioning
- Dynamic — Implementing-Partner Certified: enterprise alumni with partner certification pathways
- Demers — Private-Practice Technocrats: invitation or upgrade for senior independent specialists
Frontend & ops
- Next.js + TypeScript + Tailwind dashboard—marketplace, profiles, bids, workspaces, real-time chat
- Docker Compose for local dev; Render for cloud; Alembic + Django migrations for dual databases
Impact
A live, operational marketplace—not a prototype. AI analyses 100% of projects on submission, ranks every bid with composite scores and per-bid reasoning, and ships multi-tenant auth, WebSocket chat, payouts, documents, and support in one stack.
Tech stack
- Django 5
- Django REST Framework
- Django Channels
- FastAPI
- Claude (Anthropic)
- PostgreSQL
- Next.js
- TypeScript
- Tailwind CSS
- JWT / SimpleJWT
- Docker
- WebSockets
- Render
More selected work