Product Manager: LLM Integrations & MCP Platform - Claude LLM Integration & MCP Platform

Pixalate is building the next generation of its data platform — and this role leads it as an individual contributor. 

As Director of Product Management for LLM Integrations & MCP Platform, you will own: 

MCP Server Product Strategy

• Define the tool taxonomy — which data capabilities become MCP tools, what inputs and outputs they expose, and how they are named for natural language discoverability
• Prioritize tool development based on internal use cases (analyst workflows, research, CS) and external use cases (client integrations, developer access, agentic workflows)
• Own the product roadmap from v1 through GA, including feature gates, versioning, and deprecation policy

LLM Integration & Use Case Development

• Design, document, and validate use cases where LLM-powered clients (Claude, ChatGPT, custom agents) query Pixalate's data via MCP
• Write effective, structured tool descriptions and parameter schemas that work well as LLM-readable context — understanding how models interpret tool definitions is essential
• Partner with research and data science to prototype agentic workflows and evaluate output quality, hallucination risk, and response structure

Specification & Execution

• Write TL;DR-format PRDs: a tight problem statement, bullet-pointed requirements, clear input/output contracts, and explicit non-goals — nothing more. Long documents do not ship; concise ones do
• Champion small, shippable iterations over large releases — break every initiative into the smallest version that delivers real value and can be in users' hands quickly
• Define and maintain API contracts between the MCP server and Pixalate's underlying data APIs, including versioning, deprecation, and schema change protocols
• Run structured QA on tool outputs to ensure responses are LLM-friendly, structurally consistent, and free of schema drift

Auth, Access, and Permissioning

• Own the product model for external auth (OAuth 2.0 or API key) and capability-scoped access — defining what each client tier can query
• Collaborate with security and engineering to ensure external MCP deployments meet compliance and data governance requirements

External Developer Experience

• Own developer-facing documentation: tool reference docs, quickstart guides, example prompts, and integration guides for Claude, ChatGPT, and Cursor
• Define the self-serve onboarding flow for external clients connecting to the Pixalate MCP server
• Gather feedback from early adopters and translate it into prioritized improvements

Cross-Timezone Collaboration

• Work fluidly across US, EU, and APAC time zones — scheduling syncs, reviews, and async communication with global engineering and client-facing teams
• Own async-first communication: write clearly enough that nothing needs a follow-up call, and structure updates so stakeholders in any time zone can act on them without waiting
• Adapt availability around key global milestones — external client launches, engineering sprints, and partner reviews that span multiple geographies

This isn't boilerplate — these are real filters for this role. If they resonate, you'll thrive here. If they don't, this is probably not the right fit.
 

• SPEED OVER PERFECTION
  • ​​​​​​​We ship early and often. A v1 in users' hands beats a v2 in a doc. We bias toward momentum and correct in flight.
  • If you are waiting for the perfect spec before starting, you are already behind.
• TL;DR IS A DISCIPLINE
  • ​​​​​​​We reject long PRDs. Every product document at Pixalate follows a TL;DR format: one-line problem statement, bullet-point requirements, done.
  • Conciseness is a sign of clear thinking. Verbosity is a red flag. Bullet points are the currency of communication here.
• OUTSIDE-IN PRODUCT THINKING
  • ​​​​​​​​​​​​​​We build from the outside in — starting with the client, the analyst, the agent workflow — and work backward to the spec. Internal opinions are hypotheses until a real user validates them.
  • Candidates who can demonstrate outside-in product decisions — shaped by client calls, API usage data, or developer feedback — stand out immediately.
• APIS ARE THE PRODUCT
  • ​​​​​​​At Pixalate, the API is often the product. Understanding how clients integrate, how developers discover and call endpoints, and how data contracts affect downstream AI behavior is not a nice-to-have — it is the job.
• ​​​​​​​GLOBAL BY DEFAULT
  • ​​​​​​​​​​​​​​Pixalate operates across time zones. Async-first communication is a core competency here, not a workaround. You will be expected to work with engineers, clients, and partners across US, EU, and APAC hours.
  • Flexibility is required — not as a perk, but as a working reality. If a fixed 9-to-5 in one time zone is non-negotiable for you, this role is not the right fit.

• B.S. or B.A. in Computer Science, Computer Engineering, or a closely related technical discipline — this is a firm requirement given the depth of protocol-level and systems work involved
• 5+ years of product management experience, with at least 2–3 years owning developer tools, data APIs, AI/ML platforms, or LLM-adjacent products
• Availability and willingness to work across global time zones — US, EU, and APAC — with flexibility to accommodate distributed teams and international clients
• Deep, hands-on API experience — you have owned API products end-to-end: endpoint design, versioning, auth flows, rate limiting, developer documentation, and client onboarding. This is non-negotiable.
• Hands-on, proven experience working with LLMs — building products on top of them, evaluating model outputs, designing prompt and tool architectures (Claude, GPT-4, Gemini, or equivalent)
• Direct experience with MCP: you have set up, connected, or built an MCP server or client, and understand how tools, resources, and prompts are defined in the protocol
• Demonstrated outside-in product development — you can point to specific decisions shaped by client interviews, API usage data, developer feedback, or real usage patterns rather than internal assumptions
• Experience designing tool or function definitions for LLM consumption — you understand how model context, tool descriptions, and parameter schemas shape agent behavior and output quality
• Track record of delivering data or developer products through fast, iterative releases — not big-bang launches

• Monthly Internet Reimbursement
• Casual, Remote Work Environment
• Flexible Hours
• Opportunity for Advancement and Professional Development
• Fun Annual Team Events
• Being part of a high-performing global team that wants to win and have fun doing it