What You Will be Doing
Own the strategy and roadmap for AI strategy within R&D and AI agents specifically,
Lead the AI adoption effort technically within the product and across the organization as the company learns how to become more efficient with AI usage for R&D.
Design and build MCP servers that expose our APIs as well-described, reliably usable tools for AI clients.
Build the AI decision layer – the prompting, tool schemas, and agentic logic that let a model select the right API, pass the right parameters, and recover gracefully from errors.
Implement complete, production-grade features end to end: backend services, data models, APIs, and the AI integration on top, not prototypes.
Define how AI-accessible APIs are scoped, authenticated, and permissioned, so an agent can only ever do what it is authorized to do. Security is the product, this matters.
Build evaluation and observability for the AI layer: measure tool-selection accuracy, catch regressions, and trace agent behavior in production.
Set engineering standards and mentor others as the effort grows into a team.
Implement new features required by our cybersecurity product.
What You Need for the Role
6+ years building and shipping production backend systems, including experience leading technical initiatives, processes and teams.
5+ years experience managing engineers, hiring, running 1:1s, giving feedback, and developing people, not just tech-leading.
Strong server-side engineering in at least two of: Node.js / JavaScript, Python, Java. Able to own a feature from data model to API.
Solid PostgreSQL – schema design, query writing and optimization, working with relational data at scale.
Proven experience designing and operating APIs (HTTP/REST/GraphQL) and a working command of auth and access control: OAuth 2.0, API keys, scopes, and RBAC.
Hands-on experience building LLM-powered features in production – direct work with LLM provider APIs (Anthropic, OpenAI, Gemini, Ollama or similar), function/tool calling, token optimization, structured outputs, and prompt engineering.
Experience with agentic systems: designing tool interfaces an LLM can use reliably, multi-step agent loops, and handling failure and recovery.
Working knowledge of MCP (Model Context Protocol) – or the depth in tool-calling and AI integration to ramp on it quickly – and a clear sense of what makes a tool definition easy for a model to use well.
Experience evaluating LLM systems: building eval suites, measuring tool-call and decision accuracy, and instrumenting AI behavior with tracing and observability.
A security mindset -treat prompt injection, untrusted inputs, least-privilege design, and data exposure as first-order concerns, not afterthoughts.
Comfortable in a small, fast-moving startup – high ownership, low hand-holding, and a willingness to build processes and infrastructure from scratch when needed.
