We started Generally Intelligent because we believe that software with human-level intelligence will have a transformative impact on the world. We’re dedicated to ensuring that that impact is a positive one.
We have enough funding to last for decades, and our backers include Y Combinator, researchers from OpenAI, Lightspeed, Threshold, and many others.
This role is about creating software to support our scientific and engineering-driven approach to developing machines with human level intelligence.
About the role
You will run experiments in order to more deeply understand both the current solutions (e.g., what specific deep neural networks are learning) and the current problems (e.g., what about the training data makes it difficult to learn the patterns that seem so obvious to us). These experiments include changing the network architecture as well as the training data distribution (our data is mostly simulated).
- You will, in the course of running those experiments, spend most of your time creating reusable tools and systems required to actually run them and understand the results. Examples include visualizations for the data, changes to the simulator to create new data, and code for running large scale network architecture search across hundreds or thousands of containers.
- You will mostly pair program with other researchers and engineers. We find pair programming particularly useful because ML code needs a high level of correctness, and it’s helpful to have two minds working on a really hard problem. Plus it’s just more fun!
- You will participate in (our own unique version of) daily stand-ups, research planning, strategy discussions, paper clubs, learn-a-thons, and other activities designed to give us space to learn and plan and think.
- You are a software engineer who is passionate about understanding how the mind works. You want to build machines that truly understand the world in order to usher in a new era of abundance and flourishing for humanity.
- You do not need to have significant prior experience with machine learning (though it’s certainly a plus!) We’re very open to helping you learn to become a great machine learning engineer. We have done this for some current team members, and have a number of resources that make it very easy to get up-to-speed quickly.
- You do need to be a really good software engineer. Machine learning is 80%+ software engineering, so you should be very comfortable with Python. Other tools we use include PyTorch, docker, numpy, and wandb.
- We believe in working with the absolute best people in the world (which is why we pay top of market)
- We believe that there is something special about getting the most talented people together in one room, especially for research (which is why we’re fully in-person in SF)
- We believe that curiosity and personal growth are both extremely important (which is why you will have a $30K+ yearly budget for executive coaching, books, courses, conferences, etc)
- We believe that it is critical to not just attract great individuals, but build strong relationships between them (whichis why we have quarterly team off-sites, frequent team dinners, and other events)
- We believe in investing in resources so that you can work more effectively (which is why we have a $250K+ compute budget per person, dedicated contract engineers to support our researchers individually, why we’ve built internal tools to automate hyper-parameter optimization, etc)
- We believe in creating a culture of science and truth-seeking, where no idea is too crazy to consider and no ideas are considered holy (which is why we encourage questions in the middle of company-wide presentations and why we encourage the most junior members in a meeting to weigh in first)
- We also believe in creating a culture of psychological safety, where everyone feels able to voice their opinion without fear of judgment (which is why we have “Feelings Friday”, where we have space to share how we felt during the week, and why we have a culture manual with a bunch of small habits that help encourage safe, effective communication)
- We believe in having fun, and that being in a positive environment isn't just much more productive, it's just a better way to live (which is why we take Friday afternoons off together as a team, joke around, and join in each others’ hobbies, whether it’s D&D or hiking or meditation)
- We believe that creating software with human level intelligence comes with the responsibility to ensure that it benefits all of humanity, not just a select few (which is why we're creating a public benefit corporation, why we’re interested in safety research, and why we have internal working groups on safety).
How to apply
All submissions are reviewed by a person, so we encourage you to include a cover letter. If you have any work that you can showcase (open source code, a research paper, a side project, etc), you should certainly include it!
We try to reply either way within a week or two at most (usually much sooner).
We know that talent comes from many backgrounds, and with many different skills and preferences. That’s why we have a hiring process that gives you the ability to showcase yourself in a variety of different ways, depending on what feels best for you.