Applied AI Researcher – LLM, Memory & Agentic Systems
A role for a PhD student or someone with a research background who doesn't want to stop at the paper, but can test, prototype, and translate ideas into solutions engineering can actually use.
We're looking for an AI Researcher to help us push the intelligence of our AI systems in agentic workflows, the data map, memory, retrieval, evaluations, and context handling. You'll work mainly with the Core Team and the Datamap Team on topics that are key to how a personalized AI assistant works: how the system understands the user, how it works with their data, how it keeps context, how it evaluates the relevance of information, and how it decides what to use, when, and why. The role sits between research and product: you'll read and evaluate relevant approaches from the AI/LLM world, but above all test them on concrete problems in our system. The output of your work won't be just notes from papers, but mainly testable hypotheses, prototypes, evaluation designs, and recommendations that the Core or Datamap Team can actually use.
What you'll actually work on
LLM, RAG & retrieval – testing approaches to finding relevant context, reranking, chunking, embeddings, and working with longer context
Memory & personalization – exploring what the AI assistant should remember, when it should use memory, and how to measure whether personalization really improves the answer
Datamap & structured data – designing ways to connect user objects, attributes, and relationships with natural language and AI answers
Agentic workflows & tool-use – analyzing when the AI should use a tool, how handoffs between agents should work, and where failures occur
AI behavior evaluations – designing practical tests and metrics for the quality of answers, retrieval, memory recall, handoffs, and data handling
Prototyping – building quick experiments in Python or notebooks and preparing recommendations for engineering
What will help you succeed in this role
We are looking for a mix of research thinking and a hands-on approach. What we expect from you:
a good understanding of modern AI/ML principles, especially LLMs and NLP
Python skills at the level of research prototyping
the ability to read and critically evaluate papers, blogs, benchmarks, and technical approaches
a practical understanding of at least some of these areas: RAG, embeddings, vector search, prompt engineering, LLM evaluation, information retrieval, agentic workflows, memory/context management
the ability to translate research ideas into testable hypotheses
analytical thinking and the ability to clearly explain what you found and why it matters
a willingness to work on tasks that evolve over time and don't have a fully clear brief from the start
What you'll work with
How the hiring process works
Tell us a bit about yourself
No cover letter needed. A CV, LinkedIn, GitHub, or portfolio is enough - whatever best shows your experience.
Prescreening
A short online interview where we go through your experience, expectations, and motivation. We also introduce the team, the project, and how we work.
Practical online simulation
For some positions, you will go through a short online simulation lasting up to 6 minutes. It is designed to be quick, but still show how you approach a real situation and think in practice.
Meeting with the team
An in-person or online meeting where we go deeper into your experience, way of working, and expectations. No puzzles, no whiteboard quizzing.
Offer
If it is a mutual fit, we align on the terms and send you a fair offer.
What you get on top
Fair compensation
Transparent pay that reflects your experience and drive.
Fitness & rest
Gym in the H2O coworking building, extra vacation, and sick days.
Flexibility & Focus
Hybrid work mode and regular meeting-free Focus Days.
Technology
A cutting-edge stack, our own 3D printers, and top AI conferences.
Mobility & housing
Carsharing and the option of discounted housing right in Devín.
Urban AI hub & Devín calm
Our new office in Eurovea brings the energy of the city, a modern tech vibe, and space for fast collaboration. Devín remains our place for focus, deep work, and a reset in nature.
Reach out anyway
If you feel you can help us build Jarvis in a different way, do not hesitate. We always like to hear from great people.