Senior AI Engineer (RAG & Document Intelligence)

Overview

A venture-backed fintech startup is rebuilding the audit process through automation, artificial intelligence, and continuous assurance.

Traditional audits are often slow, manual, and heavily dependent on repetitive workflows. The company is developing an AI-first platform designed to transform auditing into a faster, more transparent, and data-driven process. The long-term vision is to create audit infrastructure that can support venture capital firms, financial institutions, and large enterprises through real-time assurance and intelligent automation.

As an early engineering hire, you will play a key role in shaping both the product and the technical foundations of the platform.

The Challenge

This role is focused on building production-grade AI systems rather than research prototypes.

Initially, you will concentrate on developing and improving the document ingestion pipeline, transforming unstructured and highly variable financial documents into structured data that can power retrieval-augmented generation (RAG) systems.

Examples of documents include:

• Multi-tab Excel workbooks with inconsistent schemas
• Scanned PDFs and legacy documents
• Financial and fund-related documents not designed for machine processing
• OCR-generated content with complex table structures

Over time, responsibilities will expand across the broader AI stack, including retrieval, evaluation frameworks, generation quality, and platform architecture.

Responsibilities

AI Engineering

• Design, build, and maintain document ingestion pipelines, including parsing, OCR, table extraction, chunking, embedding, and indexing.
• Improve retrieval and generation quality through evaluation frameworks and measurable performance metrics.
• Optimize chunking strategies, reranking approaches, and retrieval pipelines for complex document types.
• Manage model selection and trade-offs between cost, latency, and performance across multiple LLM providers.
• Develop and maintain production-grade RAG systems serving real users.

Platform & Product Contributions

• Collaborate closely with AI engineers, product teams, and stakeholders to define priorities and deliver solutions.
• Contribute to backend and platform development when required.
• Take ownership of systems and participate in solving challenges across the organization.

Required Qualifications

AI & RAG Experience

• Demonstrated experience building and maintaining production RAG systems.
• Deep understanding of ingestion pipelines, chunking strategies, embeddings, retrieval methods, and citation-based generation.
• Experience evaluating and improving retrieval quality through structured testing and metrics.

Backend Engineering

• Strong proficiency in Python.
• Experience with FastAPI or equivalent backend frameworks.
• Familiarity with asynchronous programming patterns for concurrent LLM and embedding workloads.

Databases & Vector Search

• Experience operating PostgreSQL in production environments.
• Hands-on experience with vector databases such as pgvector, Pinecone, Qdrant, or Weaviate.
• Knowledge of schema design, indexing strategies, and query optimization.

LLM Integration

• Experience integrating OpenAI, Anthropic, or similar LLM APIs.
• Familiarity with structured outputs, function calling, retry strategies, and model selection trade-offs.
• Understanding of cost and latency optimization techniques.

Document Processing

• Practical experience working with PDFs, OCR workflows, spreadsheets, and document extraction pipelines.
• Understanding of the limitations and trade-offs of OCR technologies, especially around table preservation.

Retrieval & Search

• Strong knowledge of chunking methodologies, including content-aware and fixed-size approaches.
• Understanding of reranking strategies and retrieval architectures.
• Experience with BM25, vector search, and hybrid retrieval systems.

Preferred Experience

• Experience shipping production AI products and resolving real-world RAG system challenges.
• Interest in domain-specific AI applications rather than foundational model research.
• Familiarity with evaluation-driven development practices.
• Experience diagnosing issues related to data quality, retrieval performance, and system accuracy.
• Interest in owning systems end-to-end and contributing across multiple areas of the product.

Ideal Candidate

The ideal candidate enjoys building production systems, working with messy real-world data, and taking ownership of complex technical problems. This role is best suited for engineers interested in delivering practical AI solutions and helping shape both product direction and technical architecture in an early-stage environment.