Pod Assistant Jobs in Utah (NOW HIRING)

The opportunity

CaseWorthy is the unified whole-person care platform for human services. Roughly 1,000 government and nonprofit organizations run their programs on it, and the millions of people those programs serve depend on the data and decisions that move through it every day. Two things make that platform hard to replicate: CaseWorthy CORE, our unified data foundation, and Cara, the AI grounded natively in it and acting through the applications.

This role owns both — as products. Not features buried inside an application screen. The foundation layer every CaseWorthy application is built on. The person in this seat decides how AI shows up across the entire platform, sets the standard for how it is designed and trusted, and turns a one-of-a-kind data foundation into intelligence that gives caseworkers time back. If you want to lead the pod that owns the data and AI foundation of a mission-driven platform end to end — the infrastructure, the design, and the roadmap — this is that job.

Our governing principle is non-negotiable: Cara recommends; humans decide. Every Assistant we design amplifies professional judgment. It never replaces it.

What you’ll own

Two connected pillars at the platform layer:

Cara — the AI. You own Cara as a product: its infrastructure, its family of Assistants, and its roadmap from reactive assistance to guided workflows to agentic automation. This is the majority of the role and where we most need a strong owner. You are the accountable product owner for every Assistant that ships — application teams build against your specs and standards.

CaseWorthy CORE — the data foundation. You own the unified data foundation: the lakehouse, the semantic models, and the cross-program data layer that powers reporting, analytics, and every Cara interaction. CORE is read-only by design — the single source of truth Cara is grounded in.

What you will not own is the application-side feature work — how these capabilities surface inside ClientTrack, MediSked, and ServTracker. Application product managers own that. You build the foundation and the Assistants they consume, and you define the patterns; they light them up in context. Getting that boundary right — a strong platform layer that application teams extend by configuration, not one-off forks — is central to the role.

What you’ll do

• Lead the Platform Pod. Set direction and own the operating rhythm for the pod that delivers the platform layer — the Cara, CORE, and Platform Engineering teams — driving its product, engineering, and design work to outcomes. You lead the product managers within the pod: define the PM standards, rituals, and ways of working, anchored in our Agentic Development Lifecycle (ADLC).
• Own the Cara roadmap across all three phases — reactive, guided, agentic — and the sequencing that earns trust before it expands capability.
• Design the Assistants. Write the specs: the job each Assistant does, its inputs and grounding, its autonomy settings (where a human allows, approves, or pre-authorizes an action), its acceptance criteria, and its guardrails. Every spec anchors to “Cara recommends; humans decide.”
• Drive the infrastructure conversation, in partnership. Cara engineering owns the technical infrastructure decisions — model orchestration and routing, retrieval and knowledge grounding across the CaseWorthy University knowledge base, and the evolution from templatized to dynamic querying. You bring the product and cost lens and drive the decisioning alongside them.
• Own the MCP and API contract standards. Application teams build and own their MCP servers; you define the shared contract they implement — tool schemas, auth and permission scoping, consent, and the write-boundary rules that keep “Cara recommends; humans decide” intact when agents act through the applications.
• Own the unit economics. Partner with Cloud & Data Engineering on a fully-loaded cost-per-use model and design the usage guardrails that keep AI durable at scale. That same cost number both prices Cara and scores what we build next — you own it as a product input.
• Set the responsible-AI bar. Define the evaluation, safety, explainability, and human-in-the-loop standards every Assistant clears before it ships — and hold the line on them.
• Own CORE as a product — the data foundation, semantic models, ingestion, and the analytics substrate Cara queries, including its role in statewide data-infrastructure engagements. Protect the read-only discipline of the foundation.
• Build design patterns that scale. Define reusable Assistant and data patterns that application teams extend by configuration across programs and verticals — build once, scale by configuration.
• Partner across engineering. Work with the Engineering organization on the agentic development lifecycle and with the AI Center of Excellence on shared standards.
• Prioritize in the open. Run your roadmap through the product prioritization framework, with runtime cost as a first-class input for AI work, and make your decisions visible.
• Support go-to-market. Inform pricing and packaging for AI with the cost model and readiness signals — without owning the commercial motion.

What success looks like in your first year

• A repeatable Assistant design-and-evaluation standard exists, is documented, and is used by every application team shipping AI.
• CORE is the undisputed foundation for analytics and AI across the platform, and is ready to carry statewide data-infrastructure engagements.
• The first guided-phase Assistants are specified, in build, and on a credible path — with human-in-the-loop and cost discipline built in from the start.

What you bring

• 6+ years in product management, with meaningful time in platform product management and/or AI/ML product roles. You have owned a product that other teams build on.
• Hands-on AI/ML product experience shipped to production — large language models, retrieval-augmented generation, agentic systems, evaluation, prompt and context design, and model orchestration. You have shipped AI to real users, not just prototyped it.
• Data platform fluency — lakehouses, semantic models, and analytics. Familiarity with a modern data stack (Microsoft Fabric and Power BI a plus).
• Unit-economics literacy — you can reason about and manage the cost of AI (cost-per-use, token economics, COGS) and design guardrails that keep it sustainable.
• A platform mindset — you think in contracts, reusable patterns, and configuration over forking, and you treat internal application teams as your customers.
• Exceptional spec-writing and prioritization — you turn ambiguity into crisp, testable requirements and defensible sequencing.
• Experience leading product managers — setting standards, coaching, and running the rituals that make a small PM team effective, whether as a formal manager or a pod/team lead.
• A clear point of view on responsible AI — safety, guardrails, explainability, and keeping humans in the loop.

Nice to have

• Human services, govtech, healthcare, or another regulated enterprise SaaS domain.
• Experience participating in an AI FinOps or AI evaluation / quality function.
• Experience with data sovereignty and multi-tenant data foundations.