Machine Learning Engineer

About Elicit

Elicit is the AI research assistant. We use language models to help researchers figure out what's true and make better decisions, starting with common research tasks like literature review.

What we're aiming for:

1. Elicit radically increases the amount of good reasoning in the world.
   

   • For experts, Elicit pushes the frontier forward.

   • For non-experts, Elicit makes good reasoning more affordable. People who don't have the tools, expertise, time, or mental energy to make well-reasoned decisions on their own can do so with Elicit.

2. Elicit is a scalable ML system based on human-understandable task decompositions, with supervision of process, not outcomes. This expands our collective understanding of safe AGI architectures.

Our Twitter page shows how Elicit helps researchers today. Our roadmap outlines our vision for how Elicit impacts more than research in the future.

About the role

As an ML research engineer at Elicit, you will: 

• Compose together tens to thousands of calls to language models to accomplish tasks that we can't accomplish with a single call. (One way to learn what this is like and demonstrate how you'd think about this is to go through our Factored Cognition Primer, port it to more recent language models, and submit solutions for some of the exercises.)

• Curate datasets for finetuning models, e.g. for training models to extract policy conclusions from papers

• Set up evaluation metrics that tell us what changes to our models or training setup are improvements

• Scale up semantic search from a few thousand documents to 100k+ documents

About you

To help us get there, you'll need:

• A strong software engineering background. We want to apply your experience building systems, designing architecture, and thinking about good abstractions. Elicit will need you to do much more than write scripts.
  

• Familiarity with language models (training, finetuning, evaluation), or comparable machine learning or natural language processing background (e.g. experience with information extraction, semantic search)
  

• A startup mindset. We expect to measure our impact in part by the people whose lives we improve through reasoning and models of the future. We know you care about that too. You’ll want to test lots of ideas, get feedback, and watch yourself learning and growing every day.

To get a sense for how some of us look at applications, see this thread. (The short version: Wherever we can we prefer to directly evaluate work.)

You can review a longer list of the kinds of ML-related projects you'd be working on here.

Am I a good fit?

Consider these questions:

1. How does a transformer work?

2. What is a tokenizer?

3. What is a decorator in Python?

4. What are generic types?

Strong applicants will find it easy to answer these questions.

Benefits

In addition to working on important problems as part of a happy, productive, and positive team, we also offer great benefits (with some variation based on work location):

• Flexible work environment - work from our office in Oakland or remotely

• Fully covered health, dental, vision, and life insurance for you, generous coverage for the rest of your family (FSA/HSA options, too)

• Flexible vacation policy, with a minimum recommendation of 20 days / year

• 401K with a 6% employer match

• $2,000 device budget to start, with more accumulating for each month of work

• $500 / year personal development budget

• A team administrative assistant that you can delegate personal and work tasks to

• Different speakers, tutors, and coaches to facilitate professional development

• Commuter benefits

• A relocation bonus

For all roles at Elicit, we use a data-backed compensation framework to make sure our salaries are market-competitive, equitable, and simple.

• For this role, we're targeting a range of $160,000-$200,000, depending on your level and competencies that impact the scope of your role.

• You will also be given a generous equity package with 10-year exercise window.

You can find more reasons to work with us in this thread.

Join us!