Location: Tel Aviv#HybridDriveNets is a leader in high-scale disaggregated networking solutions. Founded in 2015, DriveNets modernizes the way service providers, cloud providers and hyperscalers build networks. Supporting the largest network in the world, more than half of AT&T’s backbone traffic is running on DriveNets’ Network Cloud open disaggregated architecture. Raising $587 million in three funding rounds, DriveNets is disrupting the networking market from high-scale architecture to AI platforms, and is bringing onboard the most talented people. We are seeking people that want to make an impact on the world’s leading communication networks and are experienced in networking architecture or AI infrastructure solutions.Responsibilities and Duties•      Design and build the profiled network infrastructure that teams run high-performance LLM serving services on in production.•      Build the data-path and memory-fabric infrastructure that gives teams the primitives to implement KV cache strategies — paged attention, prefix caching, eviction policies — and hit their efficiency and latency targets.•      Provision and profile the network fabric and cluster infrastructure that inference frameworks (vLLM, TGI, TensorRT-LLM, Triton) are deployed on across GPU clusters.•      Build the scheduling and network infrastructure that exposes the throughput primitives teams need to implement batching strategies (continuous batching, dynamic batching) under SLA constraints.•      Build the compute and memory-bandwidth infrastructure profiles that give teams the headroom to evaluate and apply quantization techniques (GPTQ, AWQ, FP8, INT8) with clear production tradeoffs.•      Build network-level observability infrastructure — TTFT, TPOT, tokens/sec, GPU utilization, cache hit rates — that teams instrument their inference services against.•      Design and build the transport layer (SSE, gRPC, WebSocket) that teams use to expose real-time inference APIs.•      Build the storage and network infrastructure — sharding, format conversion, runtime configuration — that model teams use to move checkpoints to production endpoints.