Head of AI

This is a fully remote position, open to applicants in Europe.

📋 Description

• Take ownership of the AI strategy and research direction: Define the technical roadmap for our AI capabilities.

• Manage agent quality and evaluation: Develop and implement frameworks to assess improvements in our investigation agents.

• Create breakthroughs independently: Rapidly prototype new techniques, integrate them into the product, and evaluate their impact.

• Conduct fine-tuning and model experiments using real data: Oversee fine-tuning pipelines, context engineering, model migration, and cost/routing optimization based on production data.

• Direct prioritization within the AI team: Act as the decision-maker regarding which methods deserve prototyping this week.

• Lead a small team through action: Establish technical direction for the AI engineers, enhancing quality through collaboration and review.

• Collaborate with the CTO and engineering leadership: Transform the AI roadmap into deliverable capabilities.

• Engage with customers: Occasionally interact directly with customers, translating security team needs into tangible improvements in the ML pipeline.

• Establish a rapid pace: Deliver prototypes within days, not quarters.

⛳️ Requirements

• Proven hands-on technical leadership: A history of leading AI projects while actively contributing to their development.

• Experience shipping LLM/agentic systems to production: You have built and managed generative-AI systems utilized by real customers, not mere research prototypes or presentations.

• In-depth technical expertise in the LLM era: You can articulate transformer architecture, training, fine-tuning (e.g., LoRA), and inference from foundational principles.

• Developed evaluation frameworks for non-deterministic systems: You have designed and executed evaluations for multi-step, non-deterministic agents, including trajectory evaluation and LLM-as-judge methods.

• A strong pedigree with a builder's mindset: Experience at a top-tier AI organization or a high-performing AI-native startup where you elevated the technical standards rather than relying on brand reputation.

• Clear startup signal: You have worked in early-stage environments or built something from the ground up.

• Sense of pace and urgency: You deliver prototypes within days.

• Precise and concise communication: You communicate effectively and succinctly in a remote-first, English-speaking team, both in writing and verbally.

• Nice to Haves: Background in security, vulnerability management, or adversarial domains is highly preferred.

• Comfortable in customer-facing scenarios, capable of translating agent behavior and capabilities into terms understandable by security teams.

• Practical experience in model cost/routing: Demonstrated ability to reduce inference costs and migrate between models in production environments.

• Proven track record at a successful AI-first startup, scaling systems from experimentation to impactful production.

• PhD or published research in ML/AI at prestigious venues, complemented by real-world production experience.

🏝️ Benefits

• Founding-level ownership and potential for significant upside.

• Substantial equity, a position in engineering leadership, and a pathway to VP of AI as the team grows around your contributions.

• Contributing to cybersecurity as a force for good.

• Your work directly aids organizations in preventing attacks.

• Tangible impact, real customers, and immediate feedback on your contributions.

• Opportunity to build an AI-native company from the ground up.

• Backed by a well-funded Series A (Theory Ventures) with a Series B on the horizon, allowing you to set the technical standards for AI in security at scale.

• A team that will inspire you to excel.

• Founders and engineers with backgrounds at Amazon, Elastic, and Tessian. Experienced hands-on leaders who have participated in multiple acquisitions and an IPO.

• Addressing one of the toughest challenges in the field. Evaluating non-deterministic, multi-step agents against ground truth is an ongoing problem, and we have established the exploit lab and 180+ tool agent infrastructure to tackle it.