index.html

---
layout: home
---
<section class="hero-banner home-hero">
  <div class="container">
    <div class="home-hero__grid">
      <div class="home-hero__copy">
        <h1>Ordered Systems Lab</h1>
        <p>
          Welcome! Our group aims to advance the principles and practice of building next-generation dependable
          systems that run on diverse computing platforms across the spectrum.
        </p>
        <div class="home-hero__actions">
          <a class="button bg" href="{{ '/about/' | relative_url }}">About the Lab</a>
          <a class="button button-light" href="{{ '/join/' | relative_url }}">Join Us</a>
        </div>
      </div>
      <div class="home-hero__visual">
        <div class="hero-visual-card">
          <img class="img-fluid" src="{{ '/assets/img/banner/hero-banner.png' | relative_url }}" alt="OrderLab illustration">
        </div>
      </div>
    </div>
  </div>
</section>
<!--================ End banner =================-->

<!--================ Start scope section =================-->
<section class="section-margin">
  <div class="container">
    <div class="scope-panel scope-panel--single">
      <div class="scope-panel__left">
        <h2>Research Interests</h2>
        <p>
          Our research spans broadly across operating systems, distributed systems, cloud computing, mobile systems, and ML infrastructure, while specializing in reliability, 
          fault tolerance, and performance. Our work combines systems building with deep insights to address real-world challenges facing modern systems and achieve <code>ORDER</code>. 
          Our research innovations cover:
        <ul class="scope-focus-list">
          <li><strong>Foundation:</strong> Design formal reasoning techniques to ensure strong correctness guarantees in complex systems.</li>
          <li><strong>Abstraction:</strong> Introduce new abstractions and interfaces to address fundamental gaps while avoiding ad-hoc designs.</li>
          <li><strong>Analysis:</strong> Develop automated program analyses, data-driven methods, and ML techniques to better reason about system behavior.</li>
          <li><strong>Runtime:</strong> Build robust runtime mechanisms to observe, mitigate, and recover from assorted issues while enabling self-adapting systems.</li>
        </ul>
        <code>ORDER := {Observable, Reliable, Defensible, Efficient, Responsive}</code>
        </p>
      </div>
    </div>
  </div>
</section>
<!--================ End scope section =================-->

<!--================ Start news preview section =================-->
<section class="section-margin section-shell">
  <div class="container">
    <div class="section-intro text-center">
      <h2 class="section-intro__title">News</h2>
    </div>
    <ul class="home-news-list">
      <li>
        <span class="home-news-list__date">Dec 2025</span>
        <span class="home-news-list__content"><a href="https://osdi.dev/" target="_blank" rel="noopener">Yuzhuo</a> successfully defended his PhD thesis titled <b><i>"Operating System Support for Reliable Software"</i></b> and will join Google after graduation. Congratulations, Dr. Jing!</span>
      </li>
      <li>
        <span class="home-news-list__date">Jul 2025</span>
        <span class="home-news-list__content"><b><a href="https://github.com/verify-llm/TrainVerify" target="_blank" rel="noopener">TrainVerify</a></b> is accepted to <a href="https://sigops.org/s/conferences/sosp/2025/">SOSP '25</a>! TrainVerify uses equivalence-based verification to provide strong correctness guarantess for the parallelization logic of distributed LLM training.</span>
      </li>
      <li>
        <span class="home-news-list__date">Jul 2025</span>
        <span class="home-news-list__content"><b><a href="https://github.com/OrderLab/phoenix" target="_blank" rel="noopener">Phoenix</a></b> is accepted to <a href="https://sigops.org/s/conferences/sosp/2025/">SOSP '25</a>! Phoenix provides OS-level optimistic recovery and partial state preservation for high-availability software.</span>
      </li>
      <li>
        <span class="home-news-list__date">Jul 2025</span>
        <span class="home-news-list__content"><b><a href="https://github.com/OrderLab/Atropos" target="_blank" rel="noopener">Atropos</a></b> is accepted to <a href="https://sigops.org/s/conferences/sosp/2025/">SOSP '25</a>! Atropos is an application overload control framework that uses targeted cancellation to maintain tight SLOs.</span>
      </li>
      <li>
        <span class="home-news-list__date">Mar 2025</span>
        <span class="home-news-list__content"><b><a href="https://github.com/OrderLab/TrainCheck" target="_blank" rel="noopener">TrainCheck</a></b> is accepted to <a href="https://www.usenix.org/conference/osdi25">OSDI '25</a>! TrainCheck automatically infers invariants tailored for DL training and uses these invariants to proactively detect silent training errors.</span>
      </li>
    </ul>
    <div class="text-center mt-4">
      <a class="button button-light" href="{{ '/news/' | relative_url }}">View All News</a>
    </div>
  </div>
</section>
<!--================ End news preview section =================-->

<!--================ Start recent projects section =================-->
<section class="section-margin project-showcase">
  <div class="container">
    <div class="section-intro text-center">
      <h2 class="section-intro__title">Recent Projects</h2>
    </div>
    <div class="project-carousel-shell mt-4">
      <div class="owl-theme owl-carousel active_course project-carousel">
        <article class="project-card">
          <div class="project-card__media">
            <img class="project-logo-fit" src="{{ '/assets/img/project/traincheck_logo.png' | relative_url }}" alt="TrainCheck" />
          </div>
          <div class="project-card__content">
            <h3>Catching Silent Errors in Deep Learning Training</h3>
            <p>TrainCheck learns semantic invariants from sample pipelines and enforces proactive checks at runtime to catch silent training errors early.</p>
            <p class="project-badge">OSDI 2025</p>
            <div class="project-links" aria-label="TrainCheck links">
              <a class="project-link-icon" href="{{ '/paper/traincheck-osdi25-preprint.pdf' | relative_url }}" target="_blank" rel="noopener" title="Paper">
                <i class="fas fa-file-alt" aria-hidden="true"></i><span>Paper</span>
              </a>
              <a class="project-link-icon" href="https://github.com/OrderLab/TrainCheck" target="_blank" rel="noopener" title="Software">
                <i class="fab fa-github" aria-hidden="true"></i><span>Software</span>
              </a>
            </div>
          </div>
        </article>

        <article class="project-card">
          <div class="project-card__media">
            <img src="{{ '/assets/img/project/trainverify.png' | relative_url }}" alt="TrainVerify figure" />
          </div>
          <div class="project-card__content">
            <h3>Equivalence-Based Verification for Distributed Training</h3>
            <p>TrainVerify verifies the parallelization logic of LLM training to eliminate subtle correctness bugs.</p>
            <p class="project-badge">SOSP 2025</p>
            <div class="project-links" aria-label="TrainVerify links">
              <a class="project-link-icon" href="{{ '/paper/trainverify-sosp25.pdf' | relative_url }}" target="_blank" rel="noopener" title="Paper">
                <i class="fas fa-file-alt" aria-hidden="true"></i><span>Paper</span>
              </a>
              <a class="project-link-icon" href="https://github.com/verify-llm/TrainVerify" target="_blank" rel="noopener" title="Software">
                <i class="fab fa-github" aria-hidden="true"></i><span>Software</span>
              </a>
            </div>
          </div>
        </article>

        <article class="project-card">
          <div class="project-card__media">
            <img src="{{ '/assets/img/project/phoenix.png' | relative_url }}" alt="Phoenix figure" />
          </div>
          <div class="project-card__content">
            <h3>Optimistic Recovery via Partial Process State Preservation</h3>
            <p>Phoenix introduces OS-level mechanisms of partial process state preservation and optimistic recovery to improve application availability.</p>
            <p class="project-badge">SOSP 2025</p>
            <div class="project-links" aria-label="Phoenix links">
              <a class="project-link-icon" href="{{ '/paper/phoenix-sosp25.pdf' | relative_url }}" target="_blank" rel="noopener" title="Paper">
                <i class="fas fa-file-alt" aria-hidden="true"></i><span>Paper</span>
              </a>
              <a class="project-link-icon" href="https://github.com/OrderLab/phoenix" target="_blank" rel="noopener" title="Software">
                <i class="fab fa-github" aria-hidden="true"></i><span>Software</span>
              </a>
            </div>
          </div>
        </article>

        <article class="project-card">
          <div class="project-card__media">
            <img src="{{ '/assets/img/project/atropos.png' | relative_url }}" alt="Atropos figure" />
          </div>
          <div class="project-card__content">
            <h3>Mitigating Application Resource Overload with Targeted Task Cancellation</h3>
            <p>Atropos is an overload control framework that uses targeted task cancellation to reduce application resource overload.</p>
            <p class="project-badge">SOSP 2025</p>
            <div class="project-links" aria-label="Atropos links">
              <a class="project-link-icon" href="{{ '/paper/atropos-sosp25.pdf' | relative_url }}" target="_blank" rel="noopener" title="Paper">
                <i class="fas fa-file-alt" aria-hidden="true"></i><span>Paper</span>
              </a>
              <a class="project-link-icon" href="https://github.com/OrderLab/Atropos" target="_blank" rel="noopener" title="Software">
                <i class="fab fa-github" aria-hidden="true"></i><span>Software</span>
              </a>
            </div>
          </div>
        </article>

        <article class="project-card">
          <div class="project-card__media">
            <img class="project-xinda-fit" src="{{ '/assets/img/project/xinda.png' | relative_url }}" alt="Xinda figure" />
          </div>
          <div class="project-card__content">
            <h3>Enhancing Slow-Fault Tolerance in Distributed Systems</h3>
            <p>Xinda provides an automated slow-fault testing pipeline for distributed systems and a lightweight library for adaptive slow fault detection.</p>
            <p class="project-badge">NSDI 2025</p>
            <div class="project-links" aria-label="Xinda links">
              <a class="project-link-icon" href="{{ '/paper/xinda-nsdi25-preprint.pdf' | relative_url }}" target="_blank" rel="noopener" title="Paper">
                <i class="fas fa-file-alt" aria-hidden="true"></i><span>Paper</span>
              </a>
              <a class="project-link-icon" href="https://github.com/OrderLab/xinda" target="_blank" rel="noopener" title="Software">
                <i class="fab fa-github" aria-hidden="true"></i><span>Software</span>
              </a>
            </div>
          </div>
        </article>
      </div>
    </div>
  </div>
</section>
<!--================ End recent projects section =================-->


<!--================ Start blog preview section =================-->
<section class="section-margin section-shell blog-preview-section">
  <div class="container">
    <div class="section-intro text-center">
      <h2 class="section-intro__title">From the Blog</h2>
      <p class="section-intro__subtitle">Updates on lab research, milestones, and practices.</p>
    </div>
    {% assign featured_posts = site.posts | slice: 0, 2 %}
    {% if featured_posts.size > 0 %}
    <div class="blog-grid">
      {% for post in featured_posts %}
      <article class="blog-card">
        <div class="meta">{{ post.date | date: "%b %-d, %Y" }}{% if post.author %} · {{ post.author }}{% endif %}</div>
        <h3><a href="{{ post.url | relative_url }}">{{ post.title }}</a></h3>
        <p>{{ post.excerpt | strip_html | truncate: 150 }}</p>
        <a class="button button-light" href="{{ post.url | relative_url }}">Read Post</a>
      </article>
      {% endfor %}
    </div>
    {% else %}
    <div class="blog-empty">
      Blog posts will appear here once they are published. You can already visit the blog home page to start writing.
    </div>
    {% endif %}
    <div class="text-center mt-4 blog-preview-section__actions">
      <a class="button button-light" href="{{ '/blog/' | relative_url }}">View All Posts</a>
    </div>
  </div>
</section>
<!--================ End blog preview section =================-->

<!--================ Start sponsor section =================-->
<section class="sponsors_logo_area section-padding">
  <div class="container">
    <div class="section-intro text-center">
      <h2 class="section-intro__title">Sponsors</h2>
      <p class="section-intro__subtitle">We appreciate our sponsors for their funding and support, which made our research possible.</p>
    </div>
    <div class="sponsor-grid">
      <div class="sponsor-card">
        <img src="{{ '/assets/img/sponsor/nsf.jpg' | relative_url }}" alt="NSF">
      </div>
      <div class="sponsor-card">
        <img src="{{ '/assets/img/sponsor/aws.jpg' | relative_url }}" alt="AWS">
      </div>
      <div class="sponsor-card">
        <img src="{{ '/assets/img/sponsor/azure.jpg' | relative_url }}" alt="Microsoft Azure">
      </div>
      <div class="sponsor-card">
        <img src="{{ '/assets/img/sponsor/google.jpg' | relative_url }}" alt="Google">
      </div>
      <div class="sponsor-card">
        <img src="{{ '/assets/img/sponsor/facebook.png' | relative_url }}" alt="Meta">
      </div>
    </div>
  </div>
</section>
<!--================ End sponsor section =================-->