Model Behavior Architect (Function Calling & LLM Evaluation) – London / Paris | Visa + Salary Insights

📍 Location: gb

🏷 Type: Full-time

Job Intelligence

📍 More in this location:
Browse AI jobs in gb

🏷 Similar roles:
View similar Full-time jobs

🌐 Explore all jobs:
View all AI job listings

Job Overview

This role focuses on shaping how advanced large language models interact with tools, APIs, and structured systems in real-world environments. The position sits within a leading frontier AI organization :contentReference[oaicite:0]{index=0}, working directly on improving function calling, agentic reasoning, and multi-step tool orchestration. The ideal candidate is a strong mix of machine learning engineer, API systems thinker, and evaluation architect. This is a highly technical position designed for professionals who understand both model behavior and production-grade system design at scale.

📅 Job Timeline & Status

This is an active, high-priority technical hiring cycle typically associated with frontier AI labs expanding core model capability teams. Given the specialized nature of function-calling architecture work and limited global talent pool, applicants should treat this as high urgency. Early applications are more likely to be reviewed directly by research and engineering leads rather than automated screening pipelines.

🌍 Work Eligibility & Location

This role is designed for globally distributed talent, with strong support for relocation and visa processing. While Hybrid work is expected between London and Paris offices, candidates from outside Europe are still eligible if they can align with European collaboration hours. Strong asynchronous communication skills are essential due to distributed research and engineering teams across multiple countries.

💰 Salary Intelligence

This compensation band reflects high-end European AI research and applied ML roles. Total compensation is expected to be highly competitive, combining base salary with meaningful equity exposure. Given the frontier nature of the work—directly influencing model reasoning systems and tool-use reliability—the role aligns with senior-level compensation in leading AI labs.

📊 Role Breakdown

This position sits at the intersection of model evaluation engineering, API schema design, and agentic AI system optimization. A significant portion of work involves designing structured evaluation systems that test how models select tools, format arguments, and execute multi-step reasoning chains. Around 30–40% of the role focuses on building evaluation pipelines, while another 25–35% involves diagnosing failure modes such as malformed function calls, hallucinated tool usage, or incorrect parameter inference. The remaining time is spent collaborating with research scientists to refine training signals and improve model reliability in production scenarios.

Key technical responsibilities include building synthetic tool environments, defining JSON schema constraints for function calling, and implementing robust testing frameworks for agent workflows. Candidates will regularly analyze edge cases where models fail under ambiguous prompts or incomplete API specifications. Strong fluency in Python, structured outputs, and LLM evaluation frameworks is essential. Additionally, understanding probabilistic reasoning in LLMs and designing feedback loops for continuous improvement will be critical to success in this role.

🧩 Required Skills & Fit

📈 Difficulty & Competitiveness

This is a highly selective role targeting engineers who can operate at research depth while maintaining production reliability. Expect intense competition from candidates in top AI labs and infrastructure teams. Strong prior exposure to LLM internals or tool-use systems significantly increases selection probability.

🚀 Career Impact

This role offers long-term strategic value due to its proximity to core model behavior design. Engineers in this position typically transition into AI research leadership, agent system architecture, or founding roles in applied AI startups. The exposure to cutting-edge tool-use modeling significantly enhances future employability across the AI industry.

📋 Key Responsibilities

The role involves designing and refining function-calling architectures that enable LLMs to interact with external systems reliably. You will build evaluation frameworks that measure accuracy in parameter selection, schema adherence, and multi-step execution flows. A major responsibility is identifying model failure modes such as incorrect tool selection, hallucinated function names, or malformed JSON outputs, and designing mitigation strategies.

You will also collaborate closely with AI scientists to improve training datasets and synthetic environments. This includes creating simulated APIs, defining edge-case test suites, and validating model behavior under stress conditions. Strong emphasis is placed on building scalable evaluation pipelines that continuously benchmark model performance across evolving agentic tasks.

🎯 Application Strategy

🧠 Application Optimization (Adaptive)

This section is personalized by seniority.

🧠 Fit & Positioning Analysis

Evaluate your match before applying.

📅 Application Signals

Hiring momentum is high, with strong indicators of active scaling in core model behavior teams. Competition is expected to be global and intense, especially from candidates with prior LLM infrastructure experience. Because the role is niche and research-adjacent, response times may be faster for highly qualified applicants, often within days of review. Urgency is high, as similar roles in frontier AI labs tend to close quickly once pipelines are filled. Applications may close without notice.

🔗 Apply for this Job

Apply on Company Site

✅ Job Source & Verification

This listing is derived from official Mistral AI career documentation and structured role descriptions published for research and engineering hiring cycles. Source credibility is high given direct company origin. Last updated: May 31, 2026, reflecting the most recent verification cycle based on available posting information. Candidates should cross-check on the official careers page before applying to confirm availability and role continuity.