Member of Technical Staff — Environments / Evals

Remanence; Paris; Hybrid; EUR 120,000 - 180,000; Full time

Benefits offered by this job

Visa sponsorship

Relocation support

Hybrid work

Job summary

Remanence is assembling an advanced team to build adaptive AI training environments and evaluators. You will craft pipelines that generate tasks, states, and tool configurations, while filtering for validity and diversity.

You’ll design curricula that adapt to model failures and progress, and train simulators from deployment traces to model user behavior and dynamics. You will also implement evaluation harnesses to detect reward exploitation and generalization gaps, leveraging PyTorch-based

Qualifications

• Experience designing curricula and tasks for learning agents.
• Knowledge of model evaluation and reliability checks.
• Interest in synthetic data and world-model concepts.

Responsibilities

• Build pipelines that generate tasks, initial states, and configurations.
• Develop curriculum strategies based on failures and learning progress.
• Train simulators from deployment traces to model behavior.
• Create reproducible evaluation harnesses and tests.

Skills

Curriculum learning

World models

Synthetic data

Evaluation

Tools

Ray

Docker

Playwright

Gymnasium

Job description

Member of Technical Staff — Environments / Evals

Remanence is pioneering the next era of enterprise AI by building intelligent systems that learn continuously from real-world execution. We transform complex enterprise workflows and business context into dynamic, interactive environments where AI agents can safely learn, adapt, and improve. By combining high-fidelity simulation environments with state-of-the-art training loops, we build specialized models that solve long-horizon, complex tasks with unmatched reliability. We’re building the most talent-dense AI team in Europe to make this happen.

You’ll build systems that automatically synthesize tasks, construct adaptive curricula, and learn simulators of real-world deployments. Your goal is to create a continually improving source of training experience, alongside evaluations that establish whether learning transfers to real use.

What you’ll work on

• Build pipelines that generate tasks, initial states, tool configurations, and verifiers; automatically filter for validity, diversity, and difficulty.

• Develop curriculum strategies that select and generate tasks based on model failures, learning progress, and gaps in capability coverage.

• Train learned simulators from deployment traces to model user behavior, tool responses, and environment dynamics. Validate their fidelity against real interactions.

• Build reproducible evaluation harnesses and held-out tests that detect reward exploitation, simulator shortcuts, contamination, and failures to generalize.

Relevant technologies

• ML Frameworks: PyTorch, transformers, and vLLM or SGLang.

• Environment execution: Ray, Docker, OpenEnv and Gymnasium-style interfaces; Playwright for browser-based tasks.

• Evaluation and data: Inspect AI and DeepEval for evaluations and custom verifiers/graders, versioned datasets with Apache Parquet, LanceDB and DVC.

About you

You combine strong engineering with experimental judgment and curiosity about model behavior. You think from first principles about what the company needs, not just the task in front of you. Experience with synthetic data, world models, agents, curriculum learning, or evaluation is valuable.

We welcome both deep specialists and generalists who master unfamiliar domains exceptionally quickly. You don’t need prior experience across the entire stack.

What we offer

• Competitive compensation and equity.

• A fast-paced environment combining frontier research with impactful real-world applications.

• Visa sponsorship and relocation support for candidates joining us in Paris or London.

• A flexible hybrid setup, with a preference for working together in person.