Join Our Team

Lab86 is hiring AI-first engineers for its core team in Zug, Switzerland. No legacy code. No daily meetings. No reports to non-technical managers.

Core team: what you'll work on and how

  • You will work on the core of our AI technology and AI-enabled applications, with real-time code in a safe subset of Rust, and non-real-time & tools in Python.
  • You will rely heavily on AI tools & agentic workflows to exceed what is possible.
  • You will use closed- and open-source LLMs from multiple vendors both for development and inside our products.
  • Everyone fullstack. Even your manager.
  • You will not work on any task for more than a week. If it can't be done in a week - it will be abandoned or not even started. It will not get boring.
  • You will work with little/no supervision. There is no one to cover for you. No hand holding.
  • Core engineering positions are on-site.

Requirements

  • Extensive experience in AI-powered development, especially with CLI tools & agents. Using tab-autocomplete in Cursor or VS Code is <1% of what is needed.
  • Deep understanding of AI/ML concepts and underlying technology (especially LLMs and their internal operation, context management, RAG).
  • Proficiency in Python, Git, and any strongly-typed non-garbage-collected language, Linux as the main development/deployment platform.
  • Fluent English - this is the language of all internal communication and documentation.
  • You can always justify your engineering choices, but accept that sometimes there are multiple correct ways to complete a task.
  • You understand all code you commit and take responsibility for it. This sounds like a joke, but it is no longer a joke.

Would be an advantage (but it's really not required for everyone)

  • Extensive experience with Prolog, OCaml, or Haskell.
  • Experience with SAT/SMT/ASP solvers.
  • Fluent German would be helpful for integration, but is not required for the job.

Would be a disadvantage

  • Extensive experience in enterprise application development and heavy frameworks.
  • Extensive experience with closed-source / serverless platforms / microservices.
  • Extensive experience in politically-driven organizations and stepping on heads.

Hiring process

Our aim is to reduce the time wasted on both sides. There are 4 stages of hiring, only 2 of which are scheduled events:

  1. Your self-assessment goes first. To help with that, we are more verbose than usual here so that you can better judge whether we will be a good fit for each other.
  2. Filling out the form on this page includes a test task. Feel free to use any AI tools, but remember that we are interested in seeing answers augmented by your experience - we already know what all naked LLMs answer to the questions below. It is advised not to spend more than 1 hour - we have no way to enforce that, but there is no need to prepare overly complicated solutions / answers.
  3. Single interview, which typically takes no more than 2 hours. There is only 1 decision maker. You will be allowed and expected to use AI tools during most of the interview process. There will be no LeetCode on the whiteboard.
  4. Paid on-site 1-day practice, where you come to us for at least 1 full day of work + 1 day for you to look around. We cover your work for this 1 day, travel and 2 nights in a local hotel. It can be scheduled on a weekend so as not to interfere with your current job.

So, in a nutshell, there is something you should know when considering working for us:

  • No legacy code.
  • No bureaucracy. No politics and non-technical managers claiming what you achieved as theirs.
  • A single person develops every feature end-to-end. So it's all on you - both wins and losses.
  • You will not be able to focus on a small piece of the product and get comfortable there. You will have to learn and figure stuff out continuously in diverse areas. Think of it as an eternal hackathon.
  • This is a startup, and it comes with risks and opportunities.
  • There are no non-technical managers tracking hours spent per task or bugging you: "Is it done already?", "Why?"
  • No daily standups. No scrum poker.
  • If you procrastinate, you will fail. We will compare your performance under minimal oversight versus no oversight. We expect you to perform better with no oversight. If it's not the case, we won't be a good match.

Application Form

Lab86 is an agile startup and we do not have fixed payment grades. Compensation will include base salary and stock options. What are your expectations for base salary?

Python Task: *

Write a Python program that connects to any LLM provider of your choice via an OpenAI-compatible API endpoint (using mini/flash model varieties or a 30B local model). The program should generate 10 diverse Fortran solutions that calculate the sum of prime numbers from 1 to N, with variety in speed and compactness. Requirements: (1) automatically verify all solutions for correctness, (2) rank them by the metric "performance * test time". Submit only the Python generator program with your API key replaced by a placeholder. Reference answer for N=1000000 is 37550402023.

Resume (PDF) *

Law requires us to ask you: