Senior AI Engineer: Production RAG/LLM Pipelines

A19 Projects is hiring a Senior AI Engineer (contract) to build production AI features for our client, a venture-backed fintech operating across 20+ countries. You'll ship LLM-powered features that process real financial data, working alongside the client's backend, data, and product teams. This is a build role, not research: you take AI from prototype to production. What you'll build: • Production LLM pipelines: RAG, prompt engineering, output parsing, chain orchestration • AI features such as spend categorization, document extraction, anomaly detection, financial Q&A, and reconciliation • Vector search (Pinecone, Weaviate, or pgvector): ingestion, chunking, embedding selection, re-ranking • Output validation, fallback handling, and confidence scoring for financial-grade reliability • Model serving with latency SLOs, monitoring, drift detection, and retraining • Clean backend integration: API contracts, circuit breakers, graceful degradation, and human-in-the-loop review Required: • 5+ years software engineering, 3+ in production AI/ML • Shipped LLM apps on OpenAI, Anthropic, or Cohere APIs in production • RAG design: chunking, embeddings, and vector databases • Strong Python and an orchestration framework (LangChain or LlamaIndex) • Model serving (REST/gRPC), validation, latency budgeting, and monitoring • Backend fundamentals: REST, PostgreSQL, async, and cloud (AWS, GCP, or Azure) • Observability: logging, tracing, and dashboards Nice to have: • Fintech or other regulated-industry experience • Prompt evaluation and A/B testing of AI outputs • ML lifecycle tools (MLflow, Weights & Biases, Vertex AI, or SageMaker) Engagement: • Fully remote; Latin America or US-timezone overlap preferred • Full-time (30+ hrs/week), Part-time (15+ hrs/week), ongoing (6+ months) • Start: ASAP To apply, answer the screening questions and include 1-2 examples of production RAG or LLM systems you've shipped (links, repos, or a short writeup).