Skip to content

README.md

Latest commit

 

History

History
149 lines (113 loc) · 5.07 KB

README.md

File metadata and controls

149 lines (113 loc) · 5.07 KB
Raw

Evoteli Documentation

Market Intelligence Platform - Fusing computer vision, satellite imagery, and geospatial data for decision-grade signals.

Overview

Evoteli is a next-generation market intelligence platform that combines:

  • Computer Vision - Real-time analytics from edge devices (NVIDIA Jetson Orin Nano)
  • Satellite Imagery - Change detection and property analysis using Sentinel-2 and Google Earth Engine
  • Geospatial Data - Territory mapping, demographic analysis, and location intelligence

Products

Commercial Intelligence

  • LotWatch - Real-time parking, drive-thru, and curbside analytics
  • TradeZone AI - Site selection and market analysis
  • GeoPulse - Construction activity monitoring
  • SurgeRadar - Demand forecasting and competitor benchmarking
  • PermitScope - Building permit tracking

Residential Intelligence

  • HomeScope - Parcel-level property intelligence suite
  • RoofIQ - Roof condition assessment and solar suitability
  • SolarFit - Solar installation opportunity scoring
  • DrivewayPro - Driveway and hardscape condition analysis
  • StormShield - Storm damage assessment

Technology Stack

100% open-source and permissively licensed (MIT/Apache 2.0/BSD):

  • Frontend: React, TypeScript, MapLibre GL JS, Deck.gl
  • Backend: FastAPI (Python), ClickHouse, PostGIS
  • Edge: PP-YOLOE (object detection), ByteTrack (tracking), NVIDIA Jetson Orin Nano
  • Infrastructure: Kafka, Valkey, SeaweedFS, Keycloak
  • Observability: Prometheus, Apache Superset, OpenSearch, Jaeger

Cost Savings: 97%+ vs. commercial alternatives ($7,566 for 3 months vs. $220k+)

Getting Started

Quick Links

Local Development

# Clone repository
git clone https://github.com/evoteli/docs
cd docs

# Start all services with Docker Compose
docker-compose up -d

# Access services
# API Gateway: http://localhost:8000
# ClickHouse: http://localhost:8123
# PostGIS: postgresql://localhost:5432/geointel
# Superset: http://localhost:8088

Documentation Structure

docs/
├── index.mdx                      # Platform overview
├── quickstart.mdx                 # Getting started guide
├── architecture.mdx               # System architecture
├── mvp-roadmap.mdx                # 90-day MVP plan
├── tech-stack.mdx                 # Technology selection
├── commercial-safe-stack.mdx      # License compliance
├── development-plan.mdx           # Agile methodology
├── implementation-roadmap.mdx     # Week-by-week tasks
├── implementation-summary.mdx     # Implementation overview
├── products/                      # Product documentation
├── api-reference/                 # API endpoints
├── concepts/                      # Core concepts
├── implementation/                # Implementation guides
└── deployment/                    # Deployment documentation

MVP Success Criteria (Day 90)

Technical

  • ✅ 10 sites ingesting with <10s lag
  • ✅ Occupancy accuracy ±8% vs. manual counts
  • ✅ RoofIQ geometry error <5% (n≥50 parcels)
  • ✅ API p95 <800ms (cached)
  • ✅ Average quality_score ≥0.7

Operational

  • ✅ 99% uptime over last 2 weeks
  • ✅ Alerts firing with <10% false positive rate
  • ✅ Zero critical security findings

Business

  • ✅ 2 pilot customers using API
  • ✅ 1 documented case study
  • ✅ Pricing model validated

Key Features

  • Privacy-First Architecture - Edge redaction, aggregation, retention policies
  • Decision-Grade Quality - Provenance tracking, confidence intervals, quality scoring
  • API-First Design - RESTful API with OAuth 2.0, OpenAPI spec
  • Real-Time Processing - <10s end-to-end latency from edge to API
  • Scalable Infrastructure - ClickHouse for 10k+ events/sec, PostGIS for 100k+ parcels

Performance Benchmarks

Metric Target Achieved
Edge Inference <100ms 42ms (PP-YOLOE-L, TensorRT FP16)
API Latency (p95) <800ms Pending
Throughput 10k events/sec Pending
Occupancy Accuracy ±8% Pending
RoofIQ Geometry Error <5% Pending

License

All components use permissively-licensed open-source software:

  • Application code: MIT License
  • PP-YOLOE: Apache 2.0
  • ByteTrack: MIT
  • FastAPI: MIT
  • ClickHouse: Apache 2.0
  • Valkey: BSD 3-Clause

See License Analysis for complete legal review.

Support

Contributors

Evoteli is built by a team dedicated to democratizing market intelligence through open-source technology.

Ready to build the future of market intelligence. 🚀