AI Content Generation Engine

Generation module for a SaaS platform that drafts brand-aligned content across formats from a structured brief, using multiple LLM providers with caching and cost controls.

Build something similar

Built under NDA

Screens, code, and the client's name aren't public. The architecture and my role are described below; happy to walk through details on a call.

Role: Full-stack developer
Timeline: 2026
Stack: PythonFastAPICeleryRedisPostgreSQLNext.jsClaude APIOpenAI API

The challenge

The platform needed to produce long-form and short-form content that matches a brand's positioning and tone, at volume, without runaway API spend or blocking the UI while generation runs.

What I built

I built the generation pipeline as background jobs: a brief is normalized into a prompt plan, routed to the best-fit model per task, and streamed back with progress tracking. Responses are cached by prompt fingerprint, and per-workspace budgets stop overruns. Details and client are under NDA.

Key features

Multi-provider routing (Claude, OpenAI, Mistral) with fallbacks
Background generation with real-time progress
Prompt caching and per-workspace cost limits
Structured outputs validated before they reach the editor