Cross-discipline technical product management of 3 generations of P4-programmable Ethernet switch chips, with 15 SKUs in 7 packages, with bandwidths ranging from 2Tbps to 25Tbps.

Customer-facing solution architect working across business and engineering disciplines to define next gen high performance data center fabric platforms

Appropedia.org is a crowd-sourced website for sharing articles of practical wisdom that empower individuals to improve their own lives. Examples include capturing rainwater, increasing crop production, and earthquake safe construction techniques. I created the California non-profit corporation and established 501(c)(3) status, and conceived of new strategies for increasing content and readership. As of 2017, over 10,000 articles in 8 languages and over 80 million unique visits from 150 different countries.

I am working with customers and engineering to develop new product architectures that leverage Barefoot technology for applications beyond traditional switching and routing.

Architecting Machine Learning and NVMe-over-FC enhancements to Analytics product line to allow storage admins to optimize and troubleshoot their dual protocol (SCSI and NVMe) SANs on their own. Recently wrote NVMe over Fibre Channel for Dummies, ensuring customers are fully aware of Fibre Channel as the low risk, high performance fabric that offers the most straightforward adoption path for NVMe over Fabrics.

Advocated for Brocade's technology strategy, especially around Software Defined Networking (SDN) which finally enabled customers to control their own network behavior. Responsible for both IP and storage networking customers across Europe, Mid-East and Africa region. Gathered inbound product requirements while I continued my role as chair of Open Network Foundation’s (ONF's) Open Datapath Working Group. Also, drove ONF-related activities at the OpenDaylight SDN controller initiative, an open source Linux Foundation project.

While serving as Chair of ONF's Open Datapath Working Group, I led the effort to translate the dream of post-1.0 OpenFlow into a model driven reality through the development of “Table Type Patterns” (TTPs) . The ONF had taken ownership of the OpenFlow protocol, almost synonymous with Software Defined Networking (SDN) in those early days. SDN promised to allow network owners to quickly manage their own network behavior in precise detail, rather than wait for vendors and standards. OF1.0 used a simple-but-limited model that mapped easily to most chips. Later versions had fewer limitations, but broke the simple model and did not work well with real world non-programmable chips. TTPs were -- like NETCONF’s Yang models -- capable of enabling a model driven approach across a broad variety of chip architectures -- fixed, configurable and programmable, so the SDN users could have their cake and eat it too. But it came too late; the SDN focus had shifted toward P4, which is aimed at programmable chips, and less portable than TTPs. (Will P4 win? Still too soon to know. Hope so...)

VH is an NGO working in Sierra Leone villages to improve health, education, living conditions and economic security. When the founder and president of VH asked me to join the board, I took the opportunity to contribute my business experience to the organization, as well as to get first hand familiarity with the needs of NGOs on the ground in developing nations in order to help the Appropedia Foundation reach out to similar constituencies.

Under contract with a former employer, I was tasked with adapting the firm’s microprocessor trace technology (which I had originally architected years earlier) to work on ARM processors. I architected and wrote Verilog for Altera FPGA for dual lane 6.25 Gbps/ lane ARM serial trace stream.

Worked with product management teams on two efforts to expand into adjacent markets, each of which required both technology and path-to-market innovation. The later effort, implemented as an “internal start-up”, targeted enterprise file storage virtualization (in contrast to Brocade’s block storage focus) for Windows clients. The earlier effort leveraged my Rhapsody Network’s ASIC-based SAN virtualization technology to build “Data Migration Manager” (DMM), a new high performance storage migration tool and related service offering.

As Cofounder and Director of ASICs, I co-developed Rhapsody’s ASIC / hardware / software system architecture and microarchitected the ARM-based dual protocol Gigabit Ethernet / Fibre Channel port chip and cell-switching core fabric chip. I hired the ASIC team, chose the tool and ASIC vendors and managed ASIC vendor relationships. Rhapsody was acquired by Brocade in 2003, and I continued as Director of ASICs for about a year after the acquisition.

Parachuted into a management crisis. Moved it from an overstretched CEO-led engineering model to a scalable org. Drove key changes (new building, comp models) to attract talent. Reduced turnover, boosted recruiting, and created QA department. Decided what NOT to do in order to meet commit dates and rebuild damaged Intel relationship.

New to the networking domain, I started with design and architecture work, then took on management of hardware and ASIC teams. Architected and led teams building 12 x 100Mbit port chip and a “Layer 3 Forwarding Engine” capable of line rate min-sized packet forwarding up to 3 Gbps, the max bandwidth for our upcoming pizza box. Prior to that, architected MAC lookup ASIC and managed hardware and ASIC teams to develop a 12/24 port 10/100Mbit Ethernet product to compete at the low end. Beat aggressive cost targets and shipped on schedule. Nortel shipped 300k+ units for over $100M.

Managed a team and created chip/system architectures for 10/100/1000 Mbit Layer 2 and Layer 3 Ethernet switches. Retained this position after Bay was acquired by Nortel.