About the Team The Infrastructure System Lab is a hybrid research and engineering group focused on building next-generation AI-native data infrastructure. Positioned at the intersection of databases, large-scale systems, and AI, the team leads innovation in areas such as vector and multi-modal databases, infrastructure optimization through machine learning, and LLM-based tooling like NL2SQL and NL2Chart. They also develop high-performance cache systems, including multi-engine key-value stores and LLM inference KV caches. The team thrives on collaboration, with researchers and engineers working closely to take ideas from paper to prototype to production. Their work supports key products used by millions and is regularly published and deployed at scale.
About the Role We are seeking a systems researcher or engineer with deep expertise in large-scale distributed storage and caching infrastructure to design and maintain a high-performance KV cache layer for large language model (LLM) inference. This role focuses on improving latency, throughput, and cost-efficiency in transformer-based model serving by optimizing the reuse of attention key-value states and prompt embeddings. You’ll work on cutting-edge AI systems problems with real-world impact, alongside a world-class team. The role offers opportunities to publish, contribute to open-source, attend top conferences, and enjoy competitive compensation, generous research resources, and an innovation-driven culture.
Responsibilities
- Design and implement a distributed KV cache system to store and retrieve intermediate states (e.g., attention keys/values) for transformer-based LLMs across GPUs or nodes.
- Optimize low-latency access and eviction policies for caching long-context LLM inputs, token streams, and reused embeddings.
- Collaborate with inference and serving teams to integrate the cache with token streaming pipelines, batched decoding, and model parallelism.
- Develop cache consistency and synchronization protocols for multi-tenant, multi-request environments.
- Implement memory-aware sharding, eviction (e.g., windowed LRU, TTL), and replication strategies across GPUs or distributed memory backends.
- Monitor system performance and iterate on caching algorithms to reduce compute costs and response time for inference workloads.
- Evaluate and, where needed, extend open-source KV stores or build custom GPU-aware caching layers (e.g., CUDA, Triton, shared memory, RDMA).
Requirements
Minimum Qualifications
- PhD in Computer Science, Applied Mathematics, Electrical Engineering, or a related technical field.
- Strong understanding of transformer-based model internals and how KV caching affects autoregressive decoding.
- Experience with distributed systems, memory management, and low-latency serving (RPC, gRPC, CUDA-aware networking).
- Familiarity with high-performance compute environments (NVIDIA GPUs, TensorRT, Triton Inference Server).
- Proficiency in languages like C++, Rust, Go, or CUDA for systems-level development.
Preferred Qualifications
- Prior experience building inference-serving systems for LLMs (e.g., vLLM, SGLang, FasterTransformer, DeepSpeed, Hugging Face Text Generation Inference).
- Experience with memory hierarchy optimization (HBM, NUMA, NVLink) and GPU-to-GPU communication (NCCL, GDR, GDS, InfiniBand).
- Exposure to cache-aware scheduling, batching, and prefetching strategies in model serving.