About the job
As a staff software engineer for GenAI inference, you will lead the architecture, development, and optimization of the inference engine that powers Databricks Foundation Model API. You’ll bridge research advances and production demands, ensuring high throughput, low latency, and robust scaling. Your work will encompass the full GenAI inference stack: kernels, runtimes, orchestration, memory, and integration with frameworks and orchestration systems.
Responsibilities
Own and drive the architecture, design, and implementation of the inference engine, and collaborate on model-serving stack optimized for large-scale LLMs inference
Partner closely with researchers to bring new model architectures or features (sparsity, activation compression, mixture-of-experts) into the engine
Lead the end-to-end optimization for latency, throughput, memory efficiency, and hardware utilization across GPUs, and accelerators
Define and guide standards to build and maintain instrumentation, profiling, and tracing tooling to uncover bottlenecks and guide optimizations
Architect scalable routing, batching, scheduling, memory management, and dynamic loading mechanisms for inference workloads
Ensure reliability, reproducibility, and fault tolerance in the inference pipelines, including A/B launches, rollback, and model versioning
Collaborate cross-functionally on Integrating with federated, distributed inference infrastructure – orchestrate across nodes, balance load, handle communication overhead
Drive cross-team collaboration: with platform engineers, cloud infrastructure, and security/compliance teams
Represent the team externally through benchmarks, whitepapers, and open-source contributions
Qualifications
Minimum
BS/MS/PhD in Computer Science, or a related field
Strong software engineering background (6+ years or equivalent) in performance-critical systems
Proven track record of owning complex system components and driving architectural decisions end-to-end
Deep understanding of ML inference internals: attention, MLPs, recurrent modules, quantization, sparse operations, etc.
Hands-on experience with CUDA, GPU programming, and key libraries (cuBLAS, cuDNN, NCCL, etc.)
Strong background in distributed systems design, including RPC frameworks, queuing, RPC batching, sharding, memory partitioning
Demonstrated ability to uncover and solve performance bottlenecks across layers (kernel, memory, networking, scheduler)
Experience building instrumentation, tracing, and profiling tools for ML models
Ability to lead through influence - work closely with ML researchers, translate novel model ideas into production systems
Excellent communication and leadership skills, with a proactive and ownership-driven mindset
Preferred
Bonus: published research or open-source contributions in ML systems, inference optimization, or model serving