[Hiring] AI Engineer @Jeeves

Role Description

As an AI Engineer at Jeeves, you will own whole workflows from end to end, not individual features. For example, you might own everything that happens between an invoice arriving and a payment going out:

• Document extraction
• Matching
• Policy checks
• Approval routing
• Posting to the customer's ERP
• Evaluation and monitoring

You will decide how to build it, ship it, measure it, and keep improving it. You'll work closely with Product, Backend Engineering, and our customers' finance teams, but you won't wait for someone to hand you a spec.

This is a senior, hands-on role for people who are AI-native in how they work. You use coding agents and AI tools every day to move faster, and you know where they fall short. It isn't a research role, and it isn't a role for someone who wants a narrow, well-defined ticket queue.

Qualifications

• 7+ years of professional software engineering experience, including at least 2 years building and operating LLM or AI-powered systems in production
• A track record of owning a large system or workflow end to end, from scoping and design through launch and iteration, with limited direction
• Hands-on experience shipping LLM-powered applications with APIs such as Anthropic, OpenAI, or similar, including structured outputs, error handling, and evaluation
• Experience designing agentic or multi-step AI workflows (tool use, orchestration, human review steps), or RAG systems with vector databases such as pgvector, Pinecone, or Weaviate
• Strong proficiency in Python, plus solid backend fundamentals: REST APIs, PostgreSQL or similar relational databases, async patterns, and a major cloud provider (AWS, GCP, or Azure)
• Uses AI coding agents and tools (for example Claude Code, Cursor, or Codex) as a regular part of how they build, and can explain where they help and where they don't
• Experience with observability for AI systems: logging, tracing, dashboards, and quality monitoring
• Professional fluency in English, written and spoken

Requirements

• Own AI Workflows End to End
• Take a finance workflow (for example invoice-to-payment, expense review, or transaction reconciliation) and own it from problem definition through production and ongoing improvement
• Design agentic and LLM-based systems that combine extraction, retrieval, reasoning, and tool use to complete multi-step work, with clear human-in-the-loop points for high-value financial decisions
• Write the design docs, make the build-vs-buy and model choices, and set the success metrics for each workflow you own
• Talk directly with customers and internal finance users to understand where the workflow breaks today and what "done" looks like
• Build Reliable AI in Production
• Build production-grade LLM pipelines: prompt and context design, structured outputs, validation, fallbacks, and confidence scoring
• Design retrieval and RAG components (chunking, embeddings, vector search, re-ranking) where a workflow needs grounding in customer documents or policies
• Integrate AI services cleanly with Jeeves's backend: clear API contracts, retries, graceful degradation, and per-customer data isolation
• Manage cost and latency across models, choosing the right model for each step
• Measure, Monitor, and Improve
• Build evaluation sets and automated evals for every workflow you own, and use them to catch regressions when prompts, models, or data change
• Instrument AI components with logging, tracing, dashboards, and alerts so you know when quality drops before a customer does
• Keep an audit trail of AI decisions that meets the standards of a regulated financial product
• Raise the Bar for AI Engineering at Jeeves
• Set patterns, shared tooling, and practices that other engineers can build on
• Show the wider team how to use coding agents and AI tools well in day-to-day engineering
• Review peers' AI system designs and share what you learn openly

Benefits

• Remote or hybrid work options available
• Flexible schedule for office visits in São Paulo, Brazil