AI Researcher – Rising Talent

Giotto is a Swiss AI research lab, founded in 2017 by EPFL mathematicians. We build sovereign AI: efficient reasoning models and an operating system that make advanced AI run on infrastructure you choose. Efficiency makes portability possible; portability makes sovereignty real.

Our models are a growing family, from compact models that run on a single GPU to flagship models for demanding enterprise workloads. Run Giotto in our European public cloud, in a dedicated cloud in Swiss or EU data centres, or fully on-premise, from a single-GPU workstation to multi-GPU servers. GDPR-aligned and ISO/IEC 27001:2022 certified.

Role Overview

We are looking for an intellectually curious AI Researcher to improve our models’ reasoning capabilities through post-training and inference.

You will work closely with our research and engineering teams to turn ideas into experiments, understand what drives results, and translate promising findings into practical improvements.

We welcome rising talent with strong programming skills and excellent scientific and mathematical thinking, including candidates moving into AI from other technical fields. Prior LLM experience is a strong plus, but is not required.

Responsibilities

• Improve Giotto’s models by applying state-of-the-art techniques in post-training and reasoning.
• Develop and test new ideas for post-training and inference.
• Design rigorous computational experiments, analyse results, and iterate on your findings.
• Build reliable research code and tools to support experimentation and evaluation.
• Work with researchers and engineers to integrate successful approaches into our models and systems.
• Follow relevant research and assess which methods could meaningfully improve model performance and efficiency.

About You

• Intellectually curious and scientifically sharp: you understand your computational experiments inside out, question your assumptions, and iterate while staying pragmatic.
• A strong programmer: you are comfortable with Python and can turn ideas into clear, reliable code.
• Rigorous in your approach: you design meaningful experiments, investigate unexpected results, and distinguish robust improvements from noise.
• Collaborative and independent: you take ownership of your work and communicate your reasoning clearly.
• Comfortable with uncertainty: you enjoy tackling open-ended problems in a fast-moving research environment.

Strong mathematical and algorithmic thinking is a significant advantage - think LeetCode Hard and beyond. A solid grasp of LLM fundamentals is also a strong plus.

Nice to Have

• Experience with PyTorch or similar deep learning frameworks.
• Familiarity with LLMs, transformer architectures, and model evaluation.
• Experience with supervised fine-tuning, reinforcement learning, preference optimisation, or inference-time reasoning.
• Experience with distributed computing, GPU workloads, or efficient inference.
• A research background in computer science, mathematics, physics, or another quantitative field.
• Evidence of exceptional problem-solving ability through research, technical projects, competitive programming, or mathematics and science competitions.

Technologies You May Work With

Python, PyTorch, the Hugging Face ecosystem, vLLM, Ray, distributed computing, Docker, and Kubernetes.

You do not need prior experience with every tool listed.

Location & Work Style

We offer full-time employment in Switzerland.

Our hybrid model supports remote work, with the team gathering in our Swiss office for one week per month.