AI Software Intern - Training, Models, Kernel/Ops (Poland - Summer 2027)

Tenstorrent; Warszawa; On-site; PLN 36,000 - 51,000; Full time

Job summary

Tenstorrent is seeking a full-time, on-site internship in Warsaw for 13 weeks (6 months preferred) across Training, Models, and Ops/Kernels teams. You will develop and profile distributed ML training workflows, optimize models for our hardware, and build high-performance kernels on Tenstorrent devices.

Your work will span Python/C++ development in a Linux environment, with exposure to ML frameworks like PyTorch, JAX, or TensorFlow.

Qualifications

• Pursuing BS/MS/PhD in CS/CE or related field.
• Comfortable with Python and C++, and Linux development environment.
• Strong fundamentals in data structures, algorithms, and systems.
• Interest in ML workloads on accelerator hardware.

Responsibilities

• Work on distributed ML training workflows and tooling.
• Contribute to model optimization for hardware acceleration.
• Assist kernel and ops optimization for Tenstorrent hardware.
• Collaborate across Training, Models and Ops/Kernels teams.

Skills

Python

C++

Linux

Data structures

Algorithms

Computer architecture

Machine learning workflows

Education

BS/MS/PhD in CS/CE or related

Tools

Git

PyTorch

JAX

TensorFlow

Profiling tools

Job description

Tenstorrent is leading the industry on cutting-edge AI technology, revolutionizing performance expectations, ease of use, and cost efficiency. With AI redefining the computing paradigm, solutions must evolve to unify innovations in software models, compilers, platforms, networking, and semiconductors. Our diverse team of technologists have developed a high performance RISC-V CPU from scratch, and share a passion for AI and a deep desire to build the best AI platform possible. We value collaboration, curiosity, and a commitment to solving hard problems. We are growing our team and looking for contributors of all seniorities.

Overview:

This posting includes 3 unique teams on Tenstorrent's AI software stack - Training, Model and Ops/Kernels. These teams contribute to production software where correctness, scalability, and performance all matter.

This is a full-time, on-site internship for a minimum of 13 weeks; 6 months is preferred. Training and Models teams will be based out of Warsaw. Ops/Kernels team may be based in Gdańsk or Warsaw. Final team placements are matched based on candidate skill sets and current team priorities closer to your start date.

What You Might Work On

• Training: Develops and profiles distributed LLM training workflows, optimizing training engines, scaling behavior, and diagnostic tools across multi-device systems.
• Models: Optimizes ML models (LLMs, vision models, video and image generation, and other architectures) for our hardware
• Kernels/Ops: Develops high performance kernels on Tenstorrent hardware

Who You Are

• Currently pursuing a BS, MS, or PhD in Computer Science, Computer Engineering, or a related field.
• Comfortable programming in Python and C++, and working in a Linux development environment.
• Grounded in data structures, algorithms, and computer architecture or systems fundamentals.
• Interested in how machine-learning workloads are implemented, measured, and improved on accelerator hardware.
• Curious about performance and comfortable debugging unfamiliar problems across several software layers.

What We Need

• Hands-on coursework, research, or personal projects in machine learning, systems software, or distributed computing.
• Ability to reason about memory layout, parallel execution, communication, and performance tradeoffs.
• Familiarity with a machine-learning framework such as PyTorch, JAX, or TensorFlow.
• Experience using Git, writing tests, and working through unfamiliar technical problems methodically.
• Exposure to accelerator programming, low-level optimization, or multi-device systems is a plus.

What You Will Learn

• How models move from a framework to execution on a custom AI accelerator.
• How large-model training and inference use data, tensor, pipeline, sequence, context, fully sharded data, and expert parallelism to coordinate computation and communication.
• How teams bring up models, validate accuracy, and diagnose regressions.
• How LLM inference differs between prefill and decode, including KV-cache and latency-throughput tradeoffs.
• How to profile workloads from the model level down to operations, kernels, memory, and data movement.
• How to leverage agentic workflows to support software development.

This offer of employment is contingent upon the applicant being eligible to access U.S. export-controlled technology. Due to U.S. export laws, including those codified in the U.S. Export Administration Regulations (EAR), the Company is required to ensure compliance with these laws when transferring technology to nationals of certain countries (such as EAR Country Groups D:1, E1, and E2). These requirements apply to persons located in the U.S. and all countries outside the U.S. As the position offered will have direct and/or indirect access to information, systems, or technologies subject to these laws, the offer may be contingent upon your citizenship/permanent residency status or ability to obtain prior license approval from the U.S. Commerce Department or applicable federal agency. If employment is not possible due to U.S. export laws, any offer of employment will be rescinded.