Director of AI Engineering

Location

Remote - United States only.

Ottimate is building the AI-native future of accounts payable. Our platform processes millions of invoices across hundreds of enterprise customers, powered by a suite of ML models and agentic workflows. As Director of AI Engineering, you will own the full AI and ML layer of our product — from invoice understanding and vendor intelligence to our conversational AP Copilot and the next generation of autonomous AP agents.

This is a hands-on leadership role. You will spend at least half your time writing code, architecting systems, and driving technical decisions alongside your team. You will also set the AI roadmap, partner cross-functionally with Product, Data, and Platform Engineering, and manage a distributed team of 8–10 engineers across Data and ML.

We are looking for a senior technical manager or director — ideally someone who has thrived at a smaller company and is ready for a career step up into broader ownership. If you are energized by shipping real AI products, working with noisy real-world financial data, and building the systems that will define how enterprises automate AP, this role is for you.

• Architect and ship production AI/ML systems — you write code, not just review it
• Own the AI roadmap end-to-end: prioritization, trade-offs, delivery
• Set technical standards for model quality, evals, observability, and reliability
• Drive adoption of agentic coding tools to multiply team velocity
• Claude Code, Cursor, Copilot, or equivalent — measure and improve PR throughput
• Partner with Platform Engineering on infrastructure, data pipelines, and APIs

• Manage a distributed team of 8–10 engineers across Data and ML disciplines
• Hire, develop, and retain engineers at all levels; build a high-trust remote culture
• Partner with Product on roadmap sequencing and scope trade-offs
• Work directly with customer-facing teams to close feedback loops on model quality
• Communicate AI capabilities and limitations clearly to non-technical stakeholders

• Own model performance metrics and drive continuous improvement pipelines
• Build and maintain evals frameworks — regression suites, human review, A/B testing
• Oversee training data collection, curation, and labeling operations
• Manage the full ML lifecycle: experimentation, deployment, monitoring, iteration
• Define and enforce quality bars for agentic workflows entering production

• Production agentic pipelines using frontier models
• Anthropic SDK
  · OpenAI SDK
  · tool use, function calling, multi-agent orchestration
• Reliable agent loop design — planning, memory, tool execution, error recovery
• RAG pipeline design — chunking, embedding models, retrieval tuning, reranking
• Evals frameworks built from scratch — correctness, regression, semantic similarity
• Observability for production AI — tracing, cost tracking, latency, failure analysis

• Fine-tuning frontier or open-source models for domain-specific tasks
• LoRA, QLoRA, instruction tuning — not just off-the-shelf API calls
• Training data collection, curation, cleaning, and labeling at scale
• LLM inference and serving optimization
• vLLM, TGI, or equivalent
• Model selection trade-offs — cost, latency, capability, context window

• Hands-on Python — comfortable writing, reviewing, and shipping production code
• PostgreSQL — schema design, query optimization, indexing strategies
• Distributed systems — async workers, queues, retries, state machines
• Celery or similar async task frameworks is a bonus
• Public-facing API design — REST, versioning, developer experience
• MCP server development — tool-accessible APIs for AI agent integration
• AWS or cloud infrastructure — enough to own AI…