AI Trader Platform

Comprehensive scaffold for an AI-assisted options recommendation platform, covering frontend, gateway, analytics microservices, ingestion, feature engineering, and infrastructure.

Monorepo Layout

• apps/frontend – Next.js UI with layered rationale panels and guided trade flow
• services/api-gateway – NestJS edge service handling auth, rate limiting, feature flags, and routing
• services/market-data – FastAPI service serving live/synthetic quotes and option chains from Redis
• services/data-ingestion – Go-based Polygon adapter streaming into Redis, Kafka, and S3
• services/options-analytics – FastAPI service scoring liquidity and suitability for option chains
• services/signals-service – FastAPI analytics for TA, sentiment, congressional, and macro signals
• services/recommendation-engine – FastAPI hybrid rule/ML scorer producing contract picks + sizing
• services/ai-orchestrator – FastAPI RAG layer crafting layered rationales via LLM gateway
• pipelines/airflow – ETL orchestration for IV rank, liquidity, feature-store builds
• pipelines/dbt – Timescale/Postgres models for analytics features
• feature-store – Feast feature view definitions + configuration
• schemas – JSON schemas for request/response contracts
• infrastructure – Docker, Kubernetes, and Terraform bootstrap assets
• docs – Architecture notes, roadmap, and onboarding material

Getting Started

1. Install dependencies

   npm install

   Individual Python/Go services manage dependencies via pyproject.toml and go.mod.

2. Run locally with Docker Compose

   docker compose -f infrastructure/docker/docker-compose.yml up --build

3. Frontend development

   npm run dev:frontend

4. API gateway

   npm run dev:api

5. Python services

   uvicorn app.main:app --reload --port 7001  # market-data

Env Configuration

Service	Key env vars
API Gateway	AUTH_JWKS_URI, AUTH_AUDIENCE, AUTH_ISSUER, downstream URLs
Market Data	REDIS_URI, REDIS_NAMESPACE, DEFAULT_CHAIN_CONTRACTS
Data Ingestion	POLYGON_API_KEY, POLYGON_SYMBOLS, POLYGON_SYNTHETIC_ONLY, REDIS_URI, KAFKA_BROKERS, S3_BUCKET
Options Analytics	MARKET_DATA_URL (optional live chain fetch)
Signals Service	SIGNALS_DB_DSN for Postgres-backed signals
Recommendation Engine	FEAST_REPO_PATH, MLFLOW_MODEL_URI to enable feature store + model registry

Continuous Integration

A GitHub Actions workflow (.github/workflows/ci.yml) runs linting, builds, and unit tests across Node, Python, and Go stacks.

Roadmap & Next Steps

• Polygon websocket adapter automatically flips to live data when credentials are supplied.
• Market Data API exposes Redis-backed quotes/chains for downstream services.
• Recommendation engine consumes Feast features and MLflow-served models when configured.
• Signals service hydrates congressional and macro context from Postgres if DSN provided.
• See docs/roadmap.md for the expanded backlog (vector DB, orchestrator upgrades, backtesting APIs).