AI/ML engineer - MCP

• Strong software engineering. You write clean, production code and own systems end to end.

• Hands-on fluency with LLMs and agentic systems, including how they fail and how to catch it, with strong opinions about context management.

• Experience with MCP servers, agent tooling, or LLM evaluation, or the ability to ramp on them fast.

• A researcher’’'’s instinct paired with a builder’’'’s bias. You can scope a spike, learn fast, and ship the result.

• A measurement mindset. You do not call something done until you can show it works.

What we are really looking for

The ideal candidate has hands-on experience with some combination of:

• Building MCP servers, MCP tools, agent tooling, or similar context/tool orchestration systems

• Designing and running LLM/agent evaluations

• Building production AI/ML or LLM-powered software, not just prototypes

• Working with messy data and turning it into useful, measurable product capabilities

• Strong context-management instincts: knowing how LLM systems fail, how to structure inputs/tools, and how to evaluate output quality

• Customer/product-facing judgment - ability to understand a customer problem, make sense of loose requirements, and help shape the right solution

MCP experience is highly preferred, especially candidates who have built MCP servers and maintained them over time. However, strong candidates with adjacent experience in agent tooling, LLM evals, tool calling, retrieval/context systems, or applied AI infrastructure may also be worth exploring.

Must-have profile

Strong candidates will likely be:

• Senior-level engineers with strong production software engineering fundamentals

• Comfortable owning systems end to end

• Experienced with LLMs, agents, evals, MCP, tool use, or context-management systems

• Comfortable working without perfectly defined specs

• Able to translate ambiguous customer/business needs into technical direction

• Measurement-oriented - they should care deeply about whether the system actually works

• Curious, pragmatic, and biased toward shipping

Nice-to-have / differentiators

Prioritize candidates who also bring:

• Experience building MCP servers and the underlying tools behind them

• Experience creating evaluation frameworks for AI/LLM/agentic systems

• Experience with computer vision, video, creative analysis, ad tech, marketing tech, or performance marketing data

• Experience working directly with product, customers, sales, or solutions teams

• Ability to explain prior work clearly and go deep on technical decisions

• Startup or small-team experience where requirements were ambiguous and pace was fast

• Build and own MCP servers and tooling that expose our creative understanding capabilities to both internal and external services.

• Develop the evaluation systems behind those capabilities: datasets, scoring, regression tests, and dashboards that quantify quality.

• Run research spikes on open questions in creative understanding and turn what you learn into shipped capabilities.

• Design agentic workflows that turn footage and performance data into real understanding of why creative works.

What you bring