Elicit is an AI research assistant that uses language models to help researchers figure out whatâs true and make better decisions, starting with common research tasks like literature review.
What we're aiming for:
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.
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.
Visit our Twitter to learn more about how Elicit is helping researchers and making progress on our mission.
AI engineering is a new category of technical work. Its emergence is linked to the increasing availability of powerful new artificial intelligence tools: notably generative machine learning systems like large language models.
The term was suggested by swyx on the Latent Space podcast and we've written about it on our blog before.
Elicit predicted the rise of powerful ML systems even before GPT-1 was trained, and has the deepest experience of building trustworthy, capable, and transparent applications using these tools.
If you're a software engineer who relishes the more difficult parts of backend codeâlike concurrency, fault-tolerance, and distributed systemsâyou could be a great AI engineer.
Since launching the newest version of Elicit last fall, response has been strong. We introduced Elicit Plus, our monthly subscription plan, and added thousands of paying users in a matter of months as well as hundreds of thousands of new sign-ups. This has been energizing for our team, but we want to ship more useful functionality to our users even faster.
We believe that building great AI-powered products requires excellence across multiple parts of the tech stack: from frontend UX to infrastructure. But one of the crux areas is certainly how we prompt, invoke, respond to, and manage the suite of different ML models required to make Elicit work. This is what an AI engineer will be responsible for at Elicit.

Backend: Node and Python.
Frontend: Next.js and TypeScript (we expect you to be 80+% focussed on backend work, however).
We like static type checking in Python and TypeScript
All infrastructure runs in Kubernetes across a couple of clouds
We use GitHub for code reviews and CI
Consider the questions:
What's the difference between anyio, trio, and asyncio?
What does the await keyword do in JavaScript?
What is a Kubernetes pod, and how is it different from a container?
How would you manage state when using an LLM to power a conversation?
If you have a solid answer for theseâwithout reference to documentationâthen we should chat!
Backend implementation of our "living document"
We believe that user interactions with language models should be much deeper than yet more chatbot interfaces.
We wrote about our living document approach which is one way in which users can have much richer LLM-powered product experiences.
You would work on, curate, extend, and improve the backend part of that technology. This is fascinating and challenging distributed systems work.
Building Elicit into a product researchers canât live without
We ship useful, exciting features out to users on a weekly basis. Your focus will be on the code which exists between the BFF endpoints and the ML models we use.
You will work on a mix of known features / fixes, prototypes to validate ideas, and exploratory projects in between.
Our team is small, so we expect you to appreciate the user needs underlying everything you work on. You should be comfortable making decisions and trade-offs that help us fulfill usersâ needs best.
Keeping Elicitâs bar for quality high
Youâll balance shipping features in the short term with building extensible and maintainable systems.
You will be responsible for your features in production: they need to be scalable, resilient, and easy to operate.
Youâll contribute to discussions around system design, performance evaluation, and architecture.
You can view sample projects here.
Strong technical aptitude that allows you to move comfortably around our back-end tech stack, with strong computer science fundamentals.
A genuine curiosity and enthusiasm for the capabilities of language models.
A defensive, fault-first mindset.
Above-average communication skills. Elicit is a small, highly-collaborative team, so youâll frequently be tapped to communicate complex engineering topics with stakeholders across the company.
A sense of joy in what you do. One thing we love about our team at Elicit is that everybodyâs a craftsperson who cares deeply about their field.
Start building foundational context
Get to know your team, our stack, and the product roadmap.
Youâll get to know our company documentation and other supporting resources like Supporting Process, not Outcomes.
Make your first contribution to Elicit
By the end of your first week, youâll have completed your first Linear issue, have a PR merged into our monorepo, gained understanding of our CI/CD pipeline, and learned about our monitoring and logging tools.
Youâll complete your first multi-issue project
As you learn the ropes, youâre able to tackle more impactful projects, with input from domain experts where you need it.
Youâre actively improving the team
Youâll have gotten into the swing of contributing to regular team meetings and hack days, and youâve demoed something youâve worked on during a team sync.
Youâve added some documentation, how-to guides, diagrams, or other resources meant to help us and new hires in the future.
Y
elicit