Foundational ML Research Engineer - Remote & Stock Options

Talantir; Paris; Remote; EUR 100,000 - 150,000; Full time

Benefits offered by this job

Remote work

Office options: Paris, Palo Alto, Wró

Job summary

Pathway is seeking 1–2 R&D Engineers for an ambitious foundational project on attention-based models. You will train distributed models, adapt architectures, and design new experiments with a flexible GPU budget, aiming for a six-digit salary depending on profile and location.

Remote-friendly position with potential office meetings in Paris, Palo Alto, or Wroclaw. Join a pioneering AI startup focused on long-sequence and dynamic data challenges.

Qualifications

• Publications in NeurIPS/ICLR/ICML or significant conceptual/code contributions.
• Contributed to an LLM training effort with multi-GPU setups.
• Minimum 6 months in a leading ML research center (e.g., Google Brain, DeepMind, Nvidia).
• Deep learning researcher with focus on Language Models and/or RL.
• Experienced with PyTorch, Jax, or TensorFlow.
• Understanding of GPU architectures, memory design, and communication.
• Familiarity with model monitoring, git, build systems, and CI/CD.

Responsibilities

• Perform distributed model training and experiment design.
• Improve/adapt model architectures based on results.
• Design tasks and experiments to advance the project.
• Optionally oversee data preparation activities.

Skills

Fluent English

PyTorch

Jax

TensorFlow

GPU architecture

Graph algorithms

Hands-on experiments

Distributed training

Code portfolio

General research

Tools

Git

CI/CD

Build systems

Job description

Pathway is seeking 1–2 R&D Engineers for an ambitious foundational project on attention-based models. You will train distributed models, adapt architectures, and design new experiments with a flexible GPU budget, aiming for a six-digit salary depending on profile and location.

Remote-friendly position with potential office meetings in Paris, Palo Alto, or Wroclaw. Join a pioneering AI startup focused on long-sequence and dynamic data challenges.