ProjectsBrandEngine — AI-Native Marketing Analytics
Stackedtensordeployed
BrandEngine — AI-Native Marketing Analytics
A unified platform that pulls organic social and paid-ad data into one place, then layers AI sentiment, OpenAI-powered content generation, and cross-platform attribution on top. Built by merging two prior products into a single multi-tenant platform.
Principal architect — end-to-end ownership
Jan 2024 – Present
What I built
- Merged two prior products (ad-engine + brandpulse) into one multi-tenant codebase
- OpenAI-powered sentiment analysis across Meta, Google, and TikTok feeds
- ML optimization module — scikit-learn + XGBoost for ROAS prediction
- RS256 JWKS auth delegated to the identity-service (same platform)
- Prometheus + Grafana + Loki observability built in from day one
- 11 Django apps with clean multi-tenant boundaries
Hard problems
- Schema reconciliation across two inherited products without a flag day
- Keeping AI costs predictable — per-tenant budget + Claude/OpenAI adapter
- Terraform-driven infra that stays declarative while products come and go
Tech stack
PythonDjangoDRFPostgreSQLRedisCeleryOpenAINext.js 15React 19PrometheusGrafanaLokiTerraformAWS
Tags
AIMarketingMulti-tenantAnalyticsStackedtensor
Source code is not in the public domain. Happy to walk through architecture or specific modules on a call — get in touch.