AI Engineer — Agentic AI & LLM Systems

We are looking for an AI Engineer to help us build intelligent systems around LLMs, knowledge bases, retrieval, and agentic AI.

You will work on problems where the answer is rarely obvious: imperfect data, changing models, difficult retrieval problems, limited GPU resources, unreliable outputs, and systems that need to work beyond a demo.

We are looking for someone who enjoys this kind of uncertainty.

Someone who can read a paper, investigate an idea, build an experiment, look at the results, change their mind when the results disagree with them, and eventually turn the useful parts into production software.

What you'll work on

• Build and improve LLM-powered applications and knowledge-based systems
• Develop RAG pipelines and intelligent agents
• Work on retrieval, ranking, context construction, and answer generation
• Connect AI systems to tools, APIs, databases, and structured information
• Design experiments to understand model and system behavior
• Build evaluation datasets and benchmarks for AI systems
• Research new models, techniques, and papers and test whether they solve real problems
• Work with GPU-based inference and model deployment
• Improve the quality, reliability, and efficiency of AI systems
• Move ideas from research → experiment → prototype → production

You will probably touch different parts of the stack depending on the problem. We don't expect you to have every answer beforehand.

Must Have

Strong programming ability

• Strong Python skills
• Comfortable writing code that goes beyond notebooks and experiments
• Good understanding of software engineering fundamentals
• Comfortable with Git, Linux, and debugging
• Able to read and modify an unfamiliar codebase

ML fundamentals

• Solid understanding of machine learning and deep learning concepts
• Understanding of neural networks and transformer-based models
• Familiarity with model training, evaluation, and inference
• Ability to reason about model behavior rather than treating models as black boxes

Problem solving

• Strong analytical thinking
• Comfortable breaking ambiguous problems into smaller experiments
• Able to distinguish between "the model is bad" and "our experiment is bad"
• Willing to investigate a problem independently before asking for the answer
• Comfortable working with incomplete information

Research mindset
You don't need to be an academic researcher.

You do need to be able to:

• Read technical papers and documentation
• Understand an unfamiliar method quickly
• Reproduce or adapt an interesting idea
• Design a reasonable experiment
• Interpret results critically
• Explain what you learned

Curiosity
This is probably the most important requirement.

AI changes quickly. A technology you learn today may be replaced next year.

We therefore value someone who can say:

"I've never used this before, but I can figure it out."

Good to Have

• Experience building applications with LLMs
• Experience with RAG or information retrieval
• Experience with AI agents or tool-using systems
• Experience with NLP or deep learning projects
• Experience working with embeddings or reranking
• Experience evaluating ML/AI systems systematically
• Experience with PyTorch or Hugging Face
• Experience deploying models or working with GPUs
• Experience with APIs, databases, and backend services
• Experience with Docker or cloud infrastructure
• Research experience or an academic project involving ML/AI

None of these are required individually.

A strong project demonstrating that you learned most of them yourself may be more interesting to us than a CV containing every keyword.

Bonus Points

• You have built an AI system that real people actually use
• You have contributed to an open-source AI/ML project
• You have reproduced an interesting research paper
• You have trained or fine-tuned a model yourself
• You have worked with GPUs beyond simply renting one for a notebook
• You have built your own evaluation or benchmarking methodology
• You have a technical GitHub repository that demonstrates how you think
• You have written about an interesting technical problem you solved
• You have a research project where the result was not what you expected, and you investigated why

What we care about most

We are not looking for someone who knows the longest list of AI frameworks.

We are looking for someone who can take:

"We have a problem. We don't know the solution."

and turn it into:

"I've investigated it. Here are three approaches, here is what I tested, here is what happened, and here is what I think we should do next."

That ability matters more than any individual technology.

One small benchmark
Please name your CV:

YOUR_NAME__AI_ENGINEER.pdf

And there is one final instruction hidden below.

We won't tell you what it is.

If you find it, follow it.

If you don't, don't worry — but we may have already learned something about you.

U2VuZCBhbiBlbWFpbCB0byBhbGlua2toOUBnbWFpbC5jb20gd2l0aCBvbmUgQUkgcGFwZXIgeW91IGZvdW5kIGludGVyZXN0aW5nIGFuZCBhIHNob3J0IGV4cGxhbmF0aW9uIG9mIHdoeSB5b3UgZm91bmQgaXQgaW50ZXJlc3Rpbmcu