Most Loved Workplace® Certified JobStaff Software Engineer - GenAI Performance and Kernel
Most Loved Workplace® Certified JobAbout the Role
At Databricks, a Most Loved Workplace® certified employer in the Technology space, Democratizing data and AI for every organization worldwide.
<p>P-1285</p>
<h3><strong>About This Role</strong></h3>
<p>As a staff software engineer for GenAI Performance and Kernel, you will own the design, implementation, optimization, and correctness of the high-performance GPU kernels powering our GenAI inference stack. You will lead development of highly-tuned, low-level compute paths, manage trade-offs between hardware efficiency and generality, and mentor others in kernel-level performance engineering. You will work closely with ML researchers, systems engineers, and product teams to push the state-of-the-art in inference performance at scale.</p>
<h3><strong>What You Will Do</strong></h3>
<ul>
<li>Lead the design, implementation, benchmarking, and maintenance of core compute kernels (e.g. attention, MLP, softmax, layernorm, memory management) optimized for various hardware backends (GPU, accelerators)</li>
<li>Drive the performance roadmap for kernel-level improvements: vectorization, tensorization, tiling, fusion, mixed precision, sparsity, quantization, memory reuse, scheduling, auto-tuning, etc.</li>
<li>Integrate kernel optimizations with higher-level ML systems</li>
<li>Build and maintain profiling, instrumentation, and verification tooling to detect correctness, performance regressions, numerical issues, and hardware utilization gaps</li>
<li>Lead performance investigations and root-cause analysis on inference bottlenecks, e.g. memory bandwidth, cache contention, kernel launch overhead, tensor fragmentation</li>
<li>Establish coding patterns, abstractions, and frameworks to modularize kernels for reuse, cross-backend portability, and maintainability</li>
<li>Influence system architecture decisions to make kernel improvements more effective (e.g. memory layout, dataflow scheduling, kernel fusion boundaries)</li>
<li>Mentor and guide other engineers
Want to learn more about what it's like to work at Databricks? View our full profile.
Prefer to apply on the company's career site?
View Original PostingWhy This Is a Most Loved Workplace® Certified Job
Systemic Collaboration
As Databricks continues our rapid growth, we strive to maintain a culture as open and transparent as it was in Databricks’s early days. Making sure employees have regular access to our CEO and co-founders to hear directly about our vision and priorities helps our employees forge a connection between Databricks priorities and their daily work.
Positive Vision for the Future
Databricks’ culture principles are derived from an internal study of the behaviors that have led to high impact work at Databricks. These values have a direct impact on company culture and play a large part in connecting teams across the organization. We believe in being customer-obsessed, truth-seeking, operating from first principles, having a bias for action, and putting Databricks first. Our co-founders are fully involved in day-to-day operations, including business and technical reviews, working alongside our team members to ensure that we are always aligned with our values.
Alignment of Values
Databricks’ CEO and Co-founder, Ali Ghodsi leads by example with a focus on truth-seeking and first principles thinking, two of Databricks’ core values that stay true to Databricks’s origins in academia. He champions the importance of data and reason to question biases and inform decisions. Ali remains hands-on with research and development and hosts company-wide weekly all hands, with a CEO “ask me anytime” session at the beginning of every agenda. His transparency and accessibility are core to the open and collaborative culture of Databricks.
Respect
At Databricks, we cultivate an environment where all employees can bring their unique selves and are empowered to do the best work of their careers. To ensure all voices are heard and ideas are valued, we conduct annual culture and pulse surveys and host all-hands Q&As to collect feedback from all employees. We then report back to Databricks on the specific changes implemented from the feedback collected to ensure employees feel heard and that their feedback is valued. In addition, we invest in our seven Employee Resource Groups (ERGs) that offer support and employee engagement events for individuals from underrepresented backgrounds and allies.
Killer Outcomes
Databricks prioritizes offering an exceptional employee experience by investing in tools, workspaces, and community-building. Our collaboration-first, hybrid model offers flexibility with in-person touchpoints like our annual employee kickoff, weekly CEO all-hands with happy hours, and monthly team building budgets. We are especially focused on maintaining a culture of transparency, honesty, and trust, with leaders actively engaging across channels like Slack and all hands while also championing inclusion efforts through ERG sponsorship. These initiatives help our team feel connected, valued, and empowered to do their best work.
What It's Like to Work Here
Frequently Asked Questions About Working at Databricks
Common questions candidates ask about this role and Databricks's workplace
Apply for Staff Software Engineer - GenAI Performance and Kernel
Submit your application directly to the Databricks team. Let them know why you'd be a great addition to their loved place to work.
