Working Student - LLM Inference Performance

Overview

Tensordyne offers a working student position focused on large language model inference performance based in Munich, Germany. The role involves working directly with hardware and software engineering teams to analyze and optimize how LLM building blocks run on custom silicon. Participants help answer key performance questions by writing and optimizing low-level kernels, benchmarking performance on real silicon, and developing profiling tools that support compilers, runtimes, and kernel optimization.

Who can apply

Applicants must currently be enrolled in a Master's degree program in Computer Science, Electrical Engineering, Mathematics, or a related technical field. Candidates need strong programming skills in Python alongside proficiency in at least one of C, C++, or Rust.

Requirements

Applicants are required to have an understanding of computer architecture, memory systems, bandwidth, and parallel computing gained through academic coursework or personal projects. Candidates should demonstrate an interest in low-level programming, hardware performance, and optimization. Additional requirements include practical experience using AI coding agents in development workflows, an experimental mindset capable of forming hypotheses and analyzing measurements, and strong communication skills. Experience with GPU or AI accelerator programming, embedded systems, performance profiling, or LLM inference fundamentals represents an advantage.