Staff Applied Research Engineer

About this role

<div class="content-intro"><div> <div> <div class="gmail_quote"> <div> <div><span id="m_1770241969069985273m_-2746164444908759431gmail-docs-internal-guid-131e4fb0-7fff-b4e9-ff50-e8cf32449b1b">CoreWeave is The Essential Cloud for AI™. Built for pioneers by pioneers, CoreWeave delivers a platform of technology, tools, and teams that enables innovators to build and scale AI with confidence. Trusted by leading AI labs, startups, and global enterprises, CoreWeave combines superior infrastructure performance with deep technical expertise to accelerate breakthroughs and turn compute into capability. Founded in 2017, CoreWeave became a publicly traded company (Nasdaq: CRWV) in March 2025. Learn more at <a href="http://www.coreweave.com/" target="_blank" data-saferedirecturl="https://www.google.com/url?q=http://www.coreweave.com&source=gmail&ust=1762613132717000&usg=AOvVaw3D-UOhNaqEvF5BEWxjYyAU">www.coreweave.com</a>.</span></div> </div> </div> </div> </div></div><h3 class="font-claude-response-body break-words whitespace-normal" data-sourcepos="3:1-3:20;33-52"><strong>What You'll Do:</strong></h3> <p class="font-claude-response-body break-words whitespace-normal" data-sourcepos="5:1-5:358;54-411">The OpenPipe team at CoreWeave is building tools to help agents <strong>learn from experience</strong>. This is a critical step to make agents reliable enough to perform long tasks autonomously, in the same way human employees are. We're systematically identifying and solving the major bottlenecks between today's tech and those future self-improving agents. So far, we've:</p> <ul class="[li_&]:mb-0 [li_&]:mt-1 [li_&]:gap-1 [&:not(:last-child)_ul]:pb-1 [&:not(:last-child)_ol]:pb-1 list-disc flex flex-col gap-1 pl-8 mb-3" data-sourcepos="7:1-9:203;413-771"> <li class="font-claude-response-body whitespace-normal break-words pl-2" data-sourcepos="7:1-7:65;413-477">Released <a href="https://github.com/openpipe/art" target="_blank">ART</a>, the easiest library for getting started with RL.</li> <li class="font-claude-response-body whitespace-normal break-words pl-2" data-sourcepos="8:1-8:91;478-568">Developed <a href="https://news.ycombinator.com/item?id=44535078" target="_blank">RULER</a>, a general-purpose reward function that works across many diverse tasks.</li> <li class="font-claude-response-body whitespace-normal break-words pl-2" data-sourcepos="9:1-9:203;569-771">Built <a href="https://wandb.ai/site/serverless-rl/" target="_blank">Serverless RL</a>, an elegant API that gives RL practitioners full control over their data, environment and reward function while letting them outsource the headaches of managing GPU infrastructure.</li> </ul> <p class="font-claude-response-body break-words whitespace-normal" data-sourcepos="11:1-11:1047;773-1819">These releases have a theme: we're systematically tackling each major roadblock to successfully training self-improving agents. Several serious challenges remain. Building simulated environments often requires substantial human labor, and existing training methods are not data efficient enough. We're laser-focused on solving these problems and making self-improvement a reality for agent developers.</p> <p class="font-claude-response-body break-words whitespace-normal" data-sourcepos="11:1-11:1047;773-1819">In startup terms, this is a classic hard-tech bet. Our roadmap involves substantial <strong>technical risk</strong>; there are still major technical problems we're facing without a proven solution. However, there is very little <strong>market risk</strong>. We've worked closely with the teams building agents at many of the top AI-native startups as well as large enterprises. <strong>If we can build this, everyone will want it.</strong> A self improving agent that learns from experience the way a human employee would could quickly capture a large fraction of the total inference market, which is worth tens of billions of dollars today and will be worth hundreds of billions in a few years.</p> <h3 class="font-claude-response-body break-words whitespace-normal" data-sourcepos="13:1-13:20;1821-1840"><strong>About the role:</strong></h3> <p class="font-claude-response-body break-words whitespace-normal" data-sourcepos="15:1-15:1325;1842-3166">You have trained LLMs to be SOTA on specific tasks. You have opinions on whether sequence-level or token-level importance ratios are more effective. You probably shared the <a href="https://arxiv.org/abs/2510.13786" target="_blank">ScaleRL</a> paper in your group chats, and kicked off a few ablations after you read it.</p> <p class="font-claude-response-body break-words whitespace-normal" data-sourcepos="15:1-15:1325;1842-3166">This is an applied research role. You will be expected to generate and investigate research ideas towards solving the remaining obstacles to <strong>continuous learning in production</strong>. You will work with the broader OpenPipe team to validate these research directions across real customer tasks. We are very GPU rich and are ready to direct an enormous amount of compute at this effort.</p> <p class="font-claude-response-body break-words whitespace-normal"…