Distinguished Engineer at Cisco · AI infrastructure · High-performance networking

LinkedIn · GitHub

I design and build the high-performance infrastructure that supports modern AI and cloud platforms. My work combines systems software, accelerated networking, storage I/O and security — from packet-processing dataplanes to the data paths that feed distributed workloads.

I drive AI infrastructure work focused on GPU-centric systems, KV-cache data movement and the networking and storage paths required to scale distributed inference efficiently.

• AI infrastructure — high-throughput, low-latency data paths for GPU- and accelerator-driven workloads
• FD.io / VPP — vector packet processing, HostStack and software dataplanes
• Cloud networking & security — SASE, service chaining, traffic visibility and scalable architectures
• Storage and data movement — SPDK, NVMe/TCP and userspace I/O paths
• Hardware acceleration — DPUs, SmartNICs and offloaded packet processing for AI platforms

• Building faster AI infrastructure data paths across networking, storage and accelerators
• Integrating VPP HostStack/VCL with SPDK NVMe/TCP for high-performance storage networking
• Exploring stateful dataplanes and hardware offload on BlueField and other DPUs

• VPP/SPDK benchmark results — anonymized data and reproducible analysis for a 100 GbE NVMe/TCP comparison on NVIDIA BlueField, with a focus on throughput and CPU efficiency for data-intensive infrastructure.
• FD.io / VPP — open-source, high-performance vector packet-processing platform.
• Calico/VPP — accelerating encrypted Kubernetes networking with a userspace VPP dataplane.

I am an Editor for AI Infrastructure and FD.io/VPP on Medium, sharing practical work on GPU systems, KV-cache data movement, accelerated networking, DPUs and high-performance storage paths.

• SPDK Inside VPP: Accelerating NVMe/TCP on BlueField — integrating SPDK with the VPP HostStack to improve high-performance storage data paths.

• Three VPP datapaths for NVIDIA ConnectX and BlueField — comparing native RDMA-DV, DPDK mlx5 and AF_XDP for efficient packet processing.

• Getting to 40G encrypted container networking with Calico/VPP — high-throughput encrypted networking for Kubernetes workloads.

• FlowFight — high-performance, low-memory top-k spreader detection for security monitoring and anomaly detection. Published in Computer Networks.

• TupleMerge — fast online packet classification for software-defined networking. Published in IEEE/ACM Transactions on Networking.

• Distinguished Engineer at Cisco, focused on cloud security and SASE
• Founder and former CTO of Qosmos, specializing in deep packet inspection and network intelligence
• Long-term contributor and technical leader in the FD.io / VPP ecosystem, with a focus on infrastructure for demanding cloud and AI workloads

The best way to reach me is through LinkedIn. I’m particularly interested in discussions around AI infrastructure, accelerated networking, data movement, open-source dataplanes and cloud security.