Research Engineer: Inference Architect for Fast LLMs Remote

Kog; Paris; Hybride; EUR 60 000 - 90 000; Plein temps

Avantages offerts par ce poste

Direct access to AMD and NVIDIA GPUs

Creative and decision-influencing environment

Remote-friendly working model

Résumé du poste

Kog is seeking an innovative architect to design and optimize model architectures focused on inference behavior. You'll experiment with cutting-edge AI technologies in a dynamic team setting.

In this role, you'll influence our model development processes while conducting research that enhances execution speed. A collaborative environment allows your technical judgment to shape key decisions with direct impact on system evolution.

Qualifications

• Experience adapting or modifying existing model architectures.
• Understanding of communication structure and layer dependencies affecting inference behavior.
• Fluency in Transformers and MoE with depth to reason across trade-offs.

Responsabilités

• Design new model architecture variants with execution constraints as input.
• Explore inference‑aware architectural variants and discover scalable compounds.
• Own the post-training pipeline for open-weight models optimized for inference speed.
• Scale large MoE models and analyze communication patterns at inference.
• Write and present findings in research papers at top venues.

Connaissances

Model design

Architectural optimization

Post-training methods

AI problem-solving

Formation

PhD or equivalent in a relevant field

Description du poste

Kog is seeking an innovative architect to design and optimize model architectures focused on inference behavior. You'll experiment with cutting-edge AI technologies in a dynamic team setting.

In this role, you'll influence our model development processes while conducting research that enhances execution speed. A collaborative environment allows your technical judgment to shape key decisions with direct impact on system evolution.