Intern, Systems Software Engineering

Description

The Samsung Global Ecosystem Team (GOST) is a geographically distributed group of engineers who contribute software and technical leadership to open ecosystems such as the Linux kernel and QEMU, working on Samsung's newest storage and memory technologies.

The internship centres on systems software for memory and storage pools that support reinforcement-learning post-training of large language models. You will work with the team on the open-source ecosystem R&D behind these technologies and see how open-source software affects the company's business.

You will learn core Linux kernel memory management for virtual machines, how to change the kernel's memory and storage management subsystem, and how to contribute to a very large open-source project. You will also collaborate across the team, the company and the wider industry.

Please note that each candidate may submit no more than 10 applications across all open roles in any 6-month period.

Responsibilities

• Profile the RL post-training loop to find limits on memory capacity, data-transfer overhead, storage stalls and GPU idle time.
• Build one runtime feature that manages bounded memory staging, asynchronous storage writes, prefetching or capacity allocation across workers.
• Connect that feature to an existing RL framework through its training and checkpoint interfaces.
• Keep snapshot consistency, policy-version provenance, buffer ownership and durable checkpoint publication intact during asynchronous work.
• Compare memory-only, storage-only and combined pool setups on training throughput, GPU-hours, local memory use and restart performance.
• Produce correctness tests, reproducible experiments, design documents and code ready for review and possible open-source release.
• Track down the resource management problem in the post-RL workload and fix it in the host OS.
• Take on other tasks as assigned.

Requirements

• You are pursuing a Master's or PhD in Computer Science.
• You have at least one academic quarter or semester left after the internship.
• You have programmed in C, C++ or Rust, and use Python or shell scripting for experiments and automation.
• You understand operating systems, memory management, concurrency and computer architecture.
• You have developed and debugged on Linux, used Git, and explained technical results. Coursework and research projects count.
• Experience with CXL, NUMA, Device-DAX, mmap, shared-memory programming or memory allocators is preferred.
• Experience with GPU programming, CUDA, DMA or profiling memory and data-transfer performance is preferred.
• Experience with distributed systems, cache management, vllm, sglang, LLM inference or open-source development is preferred.
• You adapt your style to the situation and to the varied global norms of colleagues.
• You learn eagerly, meet problems with curiosity and resilience, and look for data to build understanding.
• You build relationships, offer help without being asked, and welcome other people's approaches.
• You explore new ideas on your own initiative and adjust quickly to change.

Conditions

You will work on site every day at the San Jose headquarters, in line with the company's Flexible Work policy.

You report to a Principal Engineer.

The pay range is $28 - $68 USD per hour for all roles at this level across all US locations and functions. Actual pay depends on work location, job-related knowledge, skills and experience.

The company is an equal opportunity employer.

Benefits

• Medical, dental and vision coverage, plus an inclusive rewards plan for physical and mental wellbeing.
• Relocation stipend if your new office is more than 50 miles from home.
• Monthly housing stipend to help with a lease for your summer internship.
• Charitable giving match and regular volunteering opportunities.
• Paid therapy sessions and on-demand wellness apps.
• Onsite café and gym, plus virtual fitness classes.
• Flexible environment so you can balance your schedule.