About the projectAt Trilagi, we're tackling one of the most persistent challenges in modern IT operations: understanding why systems fail. Our research project is building an intelligent diagnostic system that revolutionize cloud platform operations through generative AI solutions.You'll be working on cutting-edge research that bridges computational linguistics and distributed AI systems. We're developing novel approaches that address fundamental gaps in current scientific knowledge:Advanced semantic processing of heterogeneous technical dataState-of-the-art hierarchical multi-agent architectures enabling complex task orchestrationThis is genuinely novel research - we're creating the mathematical foundations and algorithms that don't yet exist in academic literature or commercial products.This isn't purely academic research. You'll see your work evolve from mathematical models through experimental validation to a working prototype tested on real data from production environments. Your research will be published in conferences, and the resulting system will be deployed commercially.You'll have the creative freedom to explore multiple approaches, the resources to test ambitious ideas, and the satisfaction of contributing both to scientific knowledge and practical solutions that teams will use every day.Your responsibilitiesLead scientific design and execution of research on adaptive data processing and multi-agent communication systemsDesign and implement novel algorithms for processing heterogeneous technical dataDevelop custom evaluation metrics adapted to diagnostic contextsConduct rigorous experimental validation of different approachesPerform comprehensive statistical analysis of research outcomesDesign hierarchical agent architectures and communication protocolsDevelop algorithms dedicated for diagnostic reasoningOptimize LLM integration and utilization patternsImplement hallucination reduction mechanismsDefine diagnostic effectiveness metrics and benchmarksEnsure model stability and reliability in production-like conditionsOptimize inference speed and computational efficiencyFine-tune integrated system componentsOur requirementsMSc or PhD in Computer Science, Mathematics, Statistics, or related fieldHands-on experience with multi-agent systems or agent frameworksUnderstanding of chunking strategies, semantic fragmentation and embedding modelsExperience with Retrieval-Augmented Generation (RAG) systemsExperience with prompt engineering and LLM fine-tuningKnowledge of vector databases (Pinecone, Weaviate or Qdrant)Proficiency in Python with scientific computing and LLM tech stack (e.g. LangChain, LangGraph, AutoGen, CrewAI, MCP, Pandas, NumPy, scikit-learn, spaCy, TensorFlow/PyTorch)Strong background in machine learning and deep learningScientific publication recordGood command of Polish and EnglishNice to haveExperience with cloud platforms (AWS, GCP, Azure)Successfully deployed LLM-based systems in real-world applicationsPublished research on LLMs, RAG systems, or multi-agent architecturesFamiliarity with IT diagnostics, observability, system logs, or technical troubleshootingDemonstrated ability to solve novel problems with no existing solutionsCross-functional skills: Bridge research and practical engineeringWhat we offerFully Remote: Work from anywhere in PolandResearch Autonomy: High degree of independence in experimental design and methodologySmall Expert Team: Direct collaboration with research leaders, minimal bureaucracyStartup Mentality: Fast decision-making, open communication, impact-focused cultureModern Infrastructure: Access to cloud computing resources, latest AI tools and frameworksCutting-edge research: Work on unsolved problems at the frontier of AI applicationsScientific impact: Co-author publications in journals or conferences