Performance Engineering Intern

Job description

<p></p>
<h1 class="wnfdnti _1ibi0s33y _1ibi0s34u" data-pm-slice="1 1 []">Performance Engineering Intern</h1>
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea"><strong>Astera Labs · San Jose, CA · Internship · On-site</strong></p>
<h2 class="wnfdnti _1ibi0s33y _1ibi0s34u">About Astera Labs</h2>
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Astera Labs builds connectivity solutions that help power modern AI infrastructure. Our products connect processors, accelerators, memory, and networks in systems where throughput, latency, bandwidth, utilization, and power efficiency directly affect customer outcomes.</p>
<h2 class="wnfdnti _1ibi0s33y _1ibi0s34u">The opportunity</h2>
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Join the Performance Engineering team and help measure, understand, and improve system performance across AI and data-center workloads. You’ll work with real platforms and representative workloads to build reproducible benchmarks, analyze results, investigate bottlenecks, and communicate what the data means.</p>
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">This internship is designed for someone who enjoys combining systems knowledge, experimentation, and data analysis: define a question, design a fair test, run it repeatedly, inspect the traces and metrics, form a hypothesis, and recommend the next engineering action. You’ll work with experienced engineers on a meaningful project involving CPUs, GPUs, memory, CXL, PCIe, Ethernet, NVMe, or related connectivity systems.</p>
<h2 class="wnfdnti _1ibi0s33y _1ibi0s34u">What you’ll do</h2>
<ul class="wnfdntf">
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Learn the platform architecture, workload, benchmark methodology, tools, and success criteria for your assigned project.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Design and execute reproducible performance experiments across hardware configurations, software stacks, workloads, and operating conditions.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Measure and analyze metrics such as throughput, latency, tail latency, IOPS, memory bandwidth, GPU utilization, power, CPU utilization, and system-level performance.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Run or support benchmark suites and workloads such as MLPerf, CXL and memory benchmarks, FIO, MLC, SPEC, database or vector-database workloads, AI inference, and custom application tests.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Explore the performance impact of variables such as GPU count, batch size, memory placement, NUMA configuration, read/write mix, interleaving mode, link topology, and workload size.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Capture and analyze system traces, PCIe or GPU data flows, memory behavior, logs, counters, and other evidence to identify bottlenecks and performance regressions.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Develop Python or shell scripts to automate test execution, collect results, extract metrics from logs, convert data to analysis-ready formats, and generate plots or reports.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Compare results across configurations, reference systems, software versions, or competitor platforms while documenting assumptions and ensuring valid comparisons.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Investigate unexpected results by reproducing the behavior, narrowing the variables, and working with system, hardware, software, validation, and product teams to identify root causes.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Help build repeatable performance test plans and methodologies that can be applied to new GPUs, systems, workloads, or product generations.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Create clear technical documentation that explains the setup, workload, metrics, results, limitations, and recommendations.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Present your project, results, and key learnings at the end of the internship.</p>
</li>
</ul>
<h2 class="wnfdnti _1ibi0s33y _1ibi0s34u">What you bring</h2>
<ul class="wnfdntf">
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Coursework, projects, research, or hands-on experience in computer science, computer engineering, electrical engineering, data science, or a related field.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Foundational understanding of computer architecture, operating systems, computer networks, memory systems, GPUs, storage, or distributed systems.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Programming or scripting experience in Python, C/C++, Bash, or another language used for automation and data analysis.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Comfort working in Linux and using command-line tools, configuration files, logs, and source control.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Curiosity about why a system performs the way it does and a methodical approach to designing experiments and interpreting evidence.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Ability to work with data, identify trends and anomalies, question assumptions, and communicate technical conclusions clearly.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Ownership, persistence, and a collaborative mindset in an environment where performance questions often require iteration across multiple layers of the stack.</p>
</li>
</ul>
<h3 class="wnfdnti _1ibi0s33y _1ibi0s34u">Helpful experience</h3>
<ul class="wnfdntf">
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Benchmarking, profiling, performance modeling, workload characterization, or system-level debugging.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">MLPerf, FIO, Intel Memory Latency Checker, Linux perf, Nsight, NCCL, SPEC, or related performance tools.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">CPU/GPU architecture, PCIe, CXL, Ethernet, RDMA, NUMA, memory bandwidth, storage, or interconnects.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Throughput, latency, IOPS, bandwidth, QPS, p99 latency, utilization, power, or performance-per-watt analysis.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Python data analysis and visualization with tools such as pandas, NumPy, Matplotlib, or comparable libraries.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Slurm or other distributed-compute environments, automated test infrastructure, or CI workflows.</p>
</li>
<li class="wnfdnte _1ibi0s33w">
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Experience presenting technical results through concise reports, dashboards, plots, or demonstrations.</p>
</li>
</ul>
<h2 class="wnfdnti _1ibi0s33y _1ibi0s34u">Internship experience and compensation</h2>
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">This is a paid, hourly U.S. internship. The applicable hourly rate is determined as part of the offer process and specified in the offer letter, depending on education level ranges from $35/hour to $55/hour. U.S. interns currently receive a $500-per-week housing/relocation stipend for the duration of the internship. Interns participate in Early Career programming, including technical and professional-development sessions, community events, and an end-of-program project presentation.</p>
<h2 class="wnfdnti _1ibi0s33y _1ibi0s34u">Equal opportunity</h2>
<p class="wnfdntu _1ibi0s3f5 _1ibi0s3ce _1ibi0s3ea">Astera Labs is committed to providing equal employment opportunities to all applicants and employees. We value diverse perspectives and consider qualified candidates without regard to legally protected characteristics.</p>
<p></p>