AI Research Engineer - AI Safety / LLMs / VLMs

Visa Sponsorship Relocation
Apply
AI Summary

European Tech Recruit is partnering with an AI safety startup to build a safety, reliability, and optimization layer for advanced AI systems. The team trains proprietary LLMs that run faster and more efficiently than open source and commercial options.

Key Highlights
Train LLMs and VLMs with frameworks such as Megatron or DeepSpeed
Build and scale Mixture of Experts architectures
Develop multimodal pipelines across text, images, and audio
Technical Skills Required
Megatron DeepSpeed Triton ZeRO FSDP Mixture of Experts LLMs VLMs
Benefits & Perks
Competitive salary and stock options
Twenty days paid vacation
Hybrid work from Paris with relocation support
Top tier medical insurance
High quality hardware and tools
Paid subscriptions for AI agents and dev tools
Twice yearly team off sites

Job Description


AI Research Engineer - AI Safety / LLMs / VLMs


We're partnered with a genuinely exciting AI safety startup focused on building the safety, reliability, and optimization layer for advanced AI systems.

Their platform uses natural language policies to define model behavior, which are automatically tested and improved at scale. Backed by eleven million dollars from leading investors, the team trains proprietary LLMs that run faster and more efficiently than open source and commercial options. This is a small, focused team where you will work on challenging problems and ship quickly.


They're seeking an AI Research Engineer to join them team on a permanent basis, working on a hybrid model at their HQ in Paris.


What You Will Do

  • Train LLMs and VLMs with frameworks such as Megatron or DeepSpeed
  • Build and scale Mixture of Experts architectures
  • Develop multimodal pipelines across text, images, and audio
  • Write custom Triton kernels to remove training bottlenecks
  • Experiment with new architectures, hyperparameters, and datasets


You Are a Great Fit If You

  • Have experience fine tuning large models
  • Understand distributed training including tensor, pipeline, and expert parallelism plus ZeRO and FSDP
  • Can write Triton kernels for attention, MLPs, or custom ops
  • Have worked on multimodal systems such as Llava style or Flamingo style models
  • Focus on stability, convergence, and hardware efficiency


Compensation and Benefits

  • Competitive salary and stock options
  • Twenty days paid vacation
  • Hybrid work from Paris with relocation support
  • Top tier medical insurance
  • High quality hardware and tools
  • Paid subscriptions for AI agents and dev tools
  • Twice yearly team off sites


Visa sponsorship and visa support is on offer, so international applicants are encouraged.


If this sounds interesting and you'd like to learn more, click the link below to apply or email me with a copy of your CV on smouland@eu-recruit.com


By applying to this role you understand that we may collect your personal data and store and process it on our systems. For more information please see our Privacy Notice (https://eu-recruit.com/about-us/privacy-notice/)


Subscribe our newsletter

New Things Will Always Update Regularly