Head of Applied AI

HEAD OF APPLIED AI

ABOUT LATENT

Latent builds clinical intelligence that helps health systems run complex pharmacy operations, giving patients access to the medications they need and helping care teams focus on serving patients.

The work spans healthcare workflows, enterprise software, and applied AI. Engineers work closely with Product and customer-facing teams to understand the problem, build the solution, and own the outcome in production.

THE ROLE

We are hiring a Head of Applied AI to lead the team building the AI capabilities that power Latent’s products. You will own our approach to production AI across complex clinical and operational workflows, including prior authorization, payor communication, clinical review, and care gap analysis, etc.

You will lead the development of Latent Agent, our agent harness, as well as the evaluation systems and shared foundations that help teams build reliable clinical AI products. You will decide when to use off-the-shelf models, invest in post-training, or take on research bets where our data and health-system relationships create an advantage.

This is both a technical and organizational leadership role. You will set strategy, guide architecture, build and develop the team, and work closely with Product, Engineering, and business leaders to turn technical capabilities into measurable outcomes for health systems and patients.

RESPONSIBILITIES

- Set the strategy and operating model for Applied AI at Latent

- Lead the team building production AI agents and intelligence systems for clinical and operational workflows

- Own Latent Agent, our evaluation framework, and the shared foundations for new AI use cases

- Establish evaluation practices that drive product, technical, and business decisions

- Make pragmatic build-versus-buy decisions across models, prompting, agent infrastructure, and post-training

- Partner with Product, Engineering, and business leaders to define priorities, metrics, and ownership

- Guide technical direction while empowering technical leads and senior engineers to own execution

- Build, hire, and develop a high-performing team with clear expectations and regular feedback

- Create predictable, visible execution without adding process for its own sake

- Own the reliability of live AI systems, including monitoring, failure handling, and service-level objectives

EXPERIENCE WE VALUE

- You have built and shipped production-grade LLM agents, not only prototypes or research systems

- You have owned live production systems, including on-call responsibilities, reliability, and service-level objectives

- You have trained or fine-tuned a model, shipped it, and measured its impact against product or business outcomes

- You have strong judgment about when to post-train a model, prompt engineer, improve the agent harness, or use an off-the-shelf capability

- You have led teams that shipped products, either as a manager or as a staff-level engineer with significant leadership responsibility

- You can move between strategy and technical detail, read design documents closely, and offer concrete guidance on architecture and tradeoffs

- You have partnered closely with Product and stakeholders outside engineering to deliver business outcomes

- You have built AI for domains where deep context, complex requirements, or operational constraints define the product

- You have made difficult performance and organizational decisions with care and clarity

- You communicate your reasoning clearly, including the options you considered, your confidence, and what would change your mind

Healthcare experience, or experience building systems that use sensitive data, is helpful but not required.

HOW WE WORK

Our teams are small by design. Everyone is expected to understand the customer problem, own their decisions, and follow their work through to measurable outcomes in production.

We learn by shipping, measuring, and improving. We treat evaluations as a core part of product development. We invest in models when the product and business need them, and we stay pragmatic about using the simplest approach that works reliably.