Local Lift

Local Lift is a data-driven lead generation platform for residential contractors. The system identifies homeowners who exhibit near-term service intent by aggregating public signals (property data, code violations, storm events, permits, etc.), scores each property, and delivers exclusive leads through a territory-based subscription model.

This repository tracks the full stack implementation, starting with the Phase 1 property universe ingestion for the Austin, TX market.

Project Structure

local-lift/
├── backend/           # FastAPI backend, ingestion pipelines, scoring engine
├── docs/              # Data source references, architecture notes
└── notebooks/         # Exploratory analysis (planned)

Getting Started (Development)

Status: Early Phase 1. Instructions will be expanded as infrastructure is added.

1. Install Python 3.11+ (3.15 target) and uv or pip for dependency management.
2. Copy .env.example (pending) to .env and set secrets when available.
3. Launch the Docker Compose stack (PostgreSQL, Redis) once defined.
4. Run ingestion pipelines from backend/src/ingestion to populate the property universe.

Roadmap

• Phase 1 — Property universe, ZIP tiering, enrichment, and data validation
• Phase 2 — Pattern discovery + scoring algorithm
• Phase 3 — Production scoring engine, territory management
• Phase 4 — Lead delivery + engagement tracking
• Phase 5 — Feedback-driven continuous learning
• Phase 6 — Web application (admin + contractor portals)

See docs/data-sources for the latest research on public data APIs.

Development Environment

# Start PostgreSQL + Redis
docker compose up -d

# Apply migrations (placeholder)
# alembic upgrade head

# Run ingestion script
python3 backend/src/ingestion/property_universe.py

Note: Python 3.15 currently lacks prebuilt wheels for pydantic-core. Install dependencies using a Python 3.12 interpreter until upstream support lands, or install Rust toolchain and set PYO3_USE_ABI3_FORWARD_COMPATIBILITY=1.