Design networks and infra for AI cloud systems.

Having been in the network team for 18 years - I have been involved in many different roles and projects. Initially focused on the datacenter portion of our network, I spent a lot of time planning capacity upgrades, working with hardware based server load balancing systems and managing a decent amount of filter-based-forwarding. One of the most common issues in the datacenter related to TCP tuning/Windowing - where I wrote documentation and created training for the SA teams to better understand this and how to get good throughput between global datacenters where RTT latency may be 30-150ms. Another common issue we built documentation and training for was related to how to troubleshoot and confirm correct server L2DSR configurations.Within a few years, my area of focus shifted into the backbone - starting with turning up peering sessions and new WAN capacity. As my backbone skills progressed, I helped to document the network and train folks on backbone protocols (BGP/ISIS/MPLS/RSVP). We used the backbone to enable L2VPN services to build a corporate network overlay, went through multiple major platform changes (Cisco 6500, then Juniper T-Series, then Juniper MX-Series). In one capacity upgrade project, we required relocating many of our existing POP cages to give us more space/power/cooling. In that project, I planned all the space/power plans, tracked hardware orders/shipments, walked through sites to qualify them, coordinated and/or performed the migration CMRs and in some cases went on-site to help with the physical build. I feel it is very important to build solid relationships with all stake holders in a network by showing that I can get my hands dirty and am not above any of the work. The best way to do this is to work side-by-side with all levels and disciplines - including on site staff, NOC staff, network operations/engineering/architecture teams as well as DNS and EDGE teams. Knowing WHO to go to is just as important as knowing WHAT to do.

Documentation and development of a long term plan to increase redundancy and reduce cost and complexity of the Blockbuter network - both corporate and retail sides. This included an entire redesign, adding site redundancy from the corporate network into the Frame/Satellite cloud for communications with the stores and a redundant DSL based VPN design with dial backup for stores that couldn't get line-of-site or rooftop access for satellite connectivity. Drove over 25% operating expenses out by solid design and contract renegotiation.

Contracted by AT&Ts to assist in the General Motors migration from the EDS WAN over to the new AT&T WAN. Acted as a member of the lifecycle management team to ensure each site cut over was successful from beginning to end.

Hired prior to relocating the corporate office. The old, antiquated MUX environment over leased lines (T-1's) and token ring needed to be fully documented and requirements gathered for the relocation. The new office design was to utilize a simple switched Ethernet LAN and a frame relay WAN, while the voice contract was renegotiated to remove the need for MUXing voice. The new network was also fully documented before I left the company and moved onto the next project.

Managed and upgraded Global WAN/LAN - including Cisco 5000 series switches, Voice/IP gateways between US and International locations to the Micom PBX and improving processes for deploying 500 new PC's on the improved network infrastructure.

Contracted by TACOM (Tank Armaments COMmand) with the Army to build and maintain their network. Responsibilities included managing the Novell infrastructure along with Windows NT DHCP servers and deploying/maintaining the Cisco network - both at TACOM and Selfridge Airforce Base.