FastAPI AI Engineer

What You’ll Build

We’re assembling the data & AI backbone for a next-generation research platform that unifies:
1. Financial-market data ingestion and enrichment
2. People & company knowledge graphs
3. Large-scale internal document chat (vector-search + retrieval-augmented generation)
4. Content-generation pipelines that publish to our Postgres-compatible data layer
5. Public web-search API orchestration
6. MCPs

The initial sprint focuses on stitching these services into a clean, scalable FastAPI backend with robust data pipelines, smart caching, and a modular plugin system for future data sources.

⸻

Core Responsibilities
• Architect & implement a FastAPI microservice that:
• Ingests & normalises bulk finance and people-data feeds
• Streams data to Supabase/Postgres and handles delta updates & observability
• Exposes REST & WebSocket endpoints with JWT auth and rate limiting
• Design the data-processing layer (ETL) with clear interface contracts, CI/CD, and proper unit / integration tests.
• Integrate third-party AI services (embeddings, RAG, agent orchestrators) and stand up a scalable vector-store workflow for internal document chat.
• Deploy on AWS using Docker & Kubernetes (EKS or similar). Configure auto-scaling, secrets management, and IaC (Terraform or CDK).
• Implement smart caching strategies for high-frequency reads and batching of slower external requests.
• Write concise technical docs (README, API reference, setup scripts) to ensure smooth hand-off and onboarding of future contributors.

⸻

Must-Have Qualifications
• 3+ years Python, with 2+ years building production FastAPI (or Starlette) services
• Proven track record designing ETL/ELT pipelines that move millions of rows daily, ideally into Postgres-family databases
• Deep familiarity with financial markets datasets (prices, fundamentals, filings, etc.) and related schema design
• Hands-on experience integrating LLM or embeddings APIs and standing up vector-search (e.g., pgvector, Pinecone, Vespa, etc.)
• Strong DevOps chops: Docker, Kubernetes, AWS networking & IAM, CI/CD workflows
• Clean API design philosophy and habit of writing well-structured, type-hinted, thoroughly tested code
• Excellent written communication; able to create clear architectural diagrams and READMEs

⸻

Nice-to-Haves
• Experience with Supabase functions / edge runtime
• Familiarity with GraphQL and WebSocket real-time streams
• Prior work on high-throughput caching layers (Redis, Cloudflare Workers, etc.)
• Comfort using cursor-based editors (Cursor, Windsurf) and AI pair-programming tools

It is also a plus if you could build AI Agent system without using slow framework like Langchain :)