Summary

Est. Compensation: 150k+/yr + Equity

Stack: Python

Much of the work we do is effectively pure software engineering. Our perspective is that even machine learning research ends up being about 90% software engineering, so even without any prior machine learning knowledge, there is plenty to contribute as a normal software engineer. Even most of our machine learning research tends towards the software engineering side of the spectrum, as we prefer to automate the types of work that academic researchers typically do (ex: tuning hyperparameters, experimenting with small variations in network architectures, etc).

Example projects

• Create an automated service for iterating on network architectures. Most of the work is in defining and implementing the search space over network (which are, at the end of the day, just code that defines a graph of computations).

• Implement new tasks in Avalon and related environments. We are constantly extending our systems to add newer, more complex tasks on which to train more capable agents.

• Optimize existing agents and models to make them lower latency and higher throughput.

• Develop improved graphing, debugging, and error handling tools to investigate the myriad ways that neural networks and agents fail.

You are

• Very comfortable writing Python.

• Familiar with PyTorch and training deep neural networks.

• Excited to work on open source code.

• Passionate about engineering best practices.

• Self-directed and independent.

• Excellent at getting things done.

Benefits

• Work directly on creating software with human-like intelligence.

• Generous compensation, equity, and benefits.

• $20K+ yearly budget for self-improvement: coaching, courses, conferences, etc.

• Actively co-create and participate in a positive, intentional team culture.

• Spend time learning, reading papers, and deeply understanding prior work.

About us

We builds AI systems that reason and code, enabling AI agents to accomplish larger goals and safely work in the real world. We train our own foundation models optimized for reasoning and prototype agents on top of these models. By using these agents extensively, we gain insights into improving both the capabilities of the underlying models and the interaction design for agents.

We aim to rekindle the dream of the personal computer, where computers become truly intelligent tools that empower us, giving us freedom, dignity, and agency to pursue the things we love.