Founding Research Engineer in the Flower Frontier Model Team at Flower

We are building the next generation of frontier AI models at Flower Labs, and we are looking for a founding Research Engineer to help shape that future from day one. Our team is pioneering decentralized learning methods that unlock sensitive data previously out of reach for large-scale training, and we are now extending that vision into a new Frontier Model Team focused on creating open-source models with superhuman capabilities across science, health, finance, and drug discovery.

What is this role?

This is a full-time senior position within a small, high-impact team that blends research and engineering expertise. The role is fully remote and open to candidates located anywhere. You will join as one of the first members of the Frontier Model Team, giving you the chance to influence both the technical direction and the culture of this new group from the ground up.

What will you do?

• Design and implement techniques across the full model development pipeline, including data curation, evaluations, pre-training, and post-training, turning promising research ideas into working systems that scale effectively.
• Run systematic experiments that combine creativity with rigorous scientific methods, using the results to guide which approaches the team adopts for its first series of frontier models.
• Collaborate closely with both research-oriented and engineering-oriented colleagues to integrate new methods into the overall training infrastructure and prepare models for open-source release and product integration.

What makes you a great fit?

You bring a deep understanding of recent architectures and training methodologies used for large language models and foundation models. You have hands-on experience with pre-training or post-training techniques such as supervised fine-tuning, reinforcement learning from human feedback, direct preference optimization, or reward modeling, and you are comfortable working with optimization methods including AdamW variants, learning rate scheduling, mixed precision, and stabilization techniques. Fluency in PyTorch or JAX is essential, along with strong experimental design skills that allow you to run controlled comparisons, ablations, and scaling experiments. You are equally at ease turning conceptual directions into runnable prototypes and collaborating across disciplines. Familiarity with Linux, Git, and Docker supports your day-to-day work, and you communicate clearly and transparently in written English. A preference will be given to candidates with post-training expertise, but we welcome any talented individual who demonstrates exceptional drive and the ability to learn quickly on the job.

What's in it for you?

You will work in a collaborative, fast-paced startup environment alongside experts who are still learning something new every day. The role offers the opportunity to contribute ideas, be heard, and influence the direction of the company across the board while helping build technologies that others will study, emulate, and build upon. As a fully remote position, you will enjoy the flexibility of working from anywhere.