Director, AI Engineering

Description

Salary Range
$201,800 – $252,300 / year

Work Policy
Hybrid, onsite 3 days a week (Princeton, NJ, San Diego, CA, or South San Francisco, CA)

Core Tech Stack
LangGraph, CrewAI, Anthropic Agents SDK, OpenAI Agents SDK, Python, PyTorch, Hugging Face Transformers

Experience Level
Director-Level (8+ years)

Job Type
Full-Time

Acadia Pharmaceuticals is hiring a Director of AI Engineering to lead the technology strategy behind its company-wide AI agent ecosystem. This is a senior leadership role that owns the underlying platform and framework agents run on, building out capabilities such as reusable agent templates, orchestration across multiple agents, an internal publishing system for agents, self-service retrieval-augmented generation (RAG) tools, registries for tools and skills, prompt management, and frameworks for evaluating and validating AI output.

About Acadia Pharmaceuticals

Acadia is a biopharmaceutical company whose AI work spans commercial, medical, R&D, and corporate teams, all operating inside strict regulatory frameworks like GxP, the NIST AI Risk Management Framework, and the EU AI Act.

What You’ll Do

• Design and maintain the platform that lets teams build, publish, version, and coordinate AI agents, using frameworks like LangGraph, CrewAI, and the Anthropic and OpenAI Agents SDKs.
• Build patterns for multiple agents with different roles and data access to work together reliably on complex, multi-step tasks.
• Create reusable templates, orchestration building blocks, and memory systems (short-term, long-term, episodic) that non-engineers can configure without deep technical knowledge.
• Set up human-in-the-loop checkpoints and clear boundaries on what agents are allowed to do autonomously, especially for higher-risk decisions.
• Lead development of a self-service RAG tool covering everything from data ingestion and chunking to embedding, vector storage, retrieval, and reranking, wrapped in a low-code interface non-technical staff can use.
• Build a searchable, versioned catalog of APIs, internal tools, MCP servers, and data connectors that agents can safely discover and call, complete with authentication and access controls.
• Stand up a prompt management system with version history, approval workflows, A/B testing, rollback options, and prompt libraries tailored to commercial, medical, and R&D teams.
• Own how AI agent performance gets measured — building rigorous test suites covering accuracy, groundedness, task success, speed, cost, and safety, with automatic detection of regressions once systems go live.
• Put multi-layered safety controls in place, covering input/output checks, content moderation, hallucination detection, and permission boundaries, aligned with company AI policy and regulatory rules.
• Treat the platform like a product — tracking how much it’s actually used, gathering feedback, and iterating to make it easier for non-engineers across Commercial, Medical Affairs, R&D, and Corporate to adopt.
• Help shape Acadia’s broader AI strategy and sit on the AI Governance Council, contributing expertise on agent risk and platform safety.
• Make sure everything built complies with global AI regulations, data privacy laws (HIPAA, GDPR), GxP requirements, and internal security standards.
• Mentor engineers and build a team culture centered on careful evaluation and responsible experimentation.

What You’ll Need

• A bachelor’s degree in Computer Science, Software Engineering, Machine Learning, AI, or a related technical field (advanced degree preferred).
• 8+ years in AI/ML or software engineering, including at least 4 years recently spent building production agentic AI systems and LLM-powered applications.
• A track record of shipping self-service tools that non-technical people actually adopt.
• Deep, hands-on expertise with at least two major agent frameworks (LangGraph, CrewAI, Anthropic Agents SDK, or OpenAI Agents SDK), including customizing them.
• Strong RAG-building experience — chunking strategy, embedding choices, vector databases, hybrid retrieval, and reranking.
• Direct experience building enterprise AI infrastructure: evaluation frameworks, guardrails, memory systems, human-in-the-loop flows, tool/skill registries, and prompt management platforms.
• Experience building or integrating Model Context Protocol (MCP) servers to expose internal capabilities to AI agents.
• Strong Python skills and familiarity with ML frameworks like PyTorch, scikit-learn, and Hugging Face Transformers, along with a solid understanding of LLM architecture.
• A habit of using AI-assisted development tools to move faster and write better code.
• Experience operating inside regulated environments and AI governance frameworks such as GxP, NIST AI RMF, or the EU AI Act.
• Biopharma industry background, ideally touching commercial, medical, clinical, or market access work and healthcare data.
• Willingness to travel domestically and internationally as needed.

Benefits

Competitive base salary plus bonus and equity (both at hire and ongoing); full medical, dental, and vision coverage; employer-paid life, disability, travel, and EAP coverage; a 401(k) with a fully vested 1:1 match up to 5%; an Employee Stock Purchase Plan with a two-year price lock; 15+ vacation days; 13-15 paid holidays including a year-end office closure; 10 paid sick days; paid parental leave; and tuition assistance.