Martin Collins Email & Phone Number

Automation, integration and managed deployments

London, United Kingdom

This big data technology company delivers innovative solutions for investment selection and oversight. It has a large production technical stack covering: Apache, Jetty/Java apps, backed with hadoop/mapreduce batch, redis, openLDAP and heavily message driven architecture utilising ActiveMQ. Development follows an agile process, meshing code review and integration tests through GitHub, Jenkins- Designed and implemented our full disk encryption method utilising TPM for key management with trusted-grub, along with custom dracut and systemd services for automatic opening of LUKS disks prior to mount- Implemented EAP-TLS 802.1x with FreeIPA based certificate management, permitting MAC based network management via MAB, EAP-TLS to assigned VLANs and unknown MACs to a guest VLANs- Migrating Jenkins static docker slave nodes to automatically scaling slave system on ec2 with Jenkins managing the full slave life-cycle to manage build queues and keep developers time to feedback as low as possible from a system perspective, using Packer to provision AMIs and gated promotion via our full integration test suite- Authored a dracut module that used any non partitioned disks (e.g. ephemeral disks on ec2) as a temporary dm_snapshot COW overlay for the base AMI root to ensure that each reboot the node would be in a clean, known state. This involved modifications to several jenkins plugins, primarily the Ec2-plugin- Streamlined the puppet configuration of a mixed role/profile pattern within the puppet code base into a simple hiera classifications with a smooth delivery of node secrets via a managed node_encrypt pipeline- Implemented/Documented a method to use the AWS APIs to spawn self-contained independent kerberized/HA Hadoop clusters in AWS for due diligence and load testing - This was was used to profile our batch scaling characteristics across both data quantity and node scaling to over 80+ AWS instances in a single placement group

London, United Kingdom

- re-engineered core services (dhcp, dns (bind), and active directory) as we as we extended to become a multi-site company- systems lead for our linux image in use on over 2000 machines in 5 geographical locations- full process from bare metal to production via pxe, kickstarts, employing extensive puppet configuration utilising puppetdb and hiera with change management via git - pulp infrastructure for yum repo management- streamlined the system image allowing it to be uploaded to cloud vendors and using provided APIs for bootstrapping machines to boot and enable burstable resourcing when required (upto several hundred cloud compute nodes)- facilitate virtually moving render-resources between site-configurations that use local resources where available for internal remote-rendering solutions via custom DHCP options/dhclient handlers- implemented nagios infrastructure utilising puppet exported-resources to automatically add monitoring of services- storage configuration, implementation and monitoring across HDS HNAS/Bluearc and Solaris/Illumos based ZFS

London, United Kingdom

- migrated the 3d department of 30 artists and ~50 render nodes from linux to windows- implemented a FOG network-based install for windows- with WPKG infrastructure to install software and keep systems up-to-date across across 150 workstations, render nodes, production laptops in multiple geographic locations to consolidate working practices- migrated away from our own AD infrastructure to that of our parent-companies to consolidate and progress single-sign-on throughout our infrastructure.- added OpenLDAP servers with synchronisation from AD using lsc to maintain consistency between linux based uids/gids/group membership for our GPFS file servers (and a number of linux clients) and windows clients as AD Schema extensions weren't permitted by our parent company- streamlined archive and backup procedures using IBM Tivoli Storage Manager (TSM) via a collection of bash scripts to allow junior staff to do the majority of a large daily turnover of archive/restore requests based on production schedules.

This role was predominantly DevOps:Ops:- maintained the online-web servers hosting image content and e-commerce- maintained internal storage/servers/workstations - aided in designing the image workflow from photo shoot -> website- validating PCI compliance in conjunction with external vulnerability scanning providersDev:- website team lead- wrote much of the site in ActionScript/PHP/MySQL- including writing full specifications and implementing a complex rights-managed licensing system with consolidation rules to choose the best pricing for clients whilst maintaining discount attribution between multiple purchasing parties (client and art-buyer)- built a dynamic ActionScript framework to allow xml driven content using light-weight dynamic components- utilising makefiles for command line mxmlc compiling to build the distributables- AS debugging and profiling

Maintained server and client systems for a small CGI company predominantly working with client's CAD data for photo-realistic car brocures- part of the same group as Moofe above- implemented small render farm using tractor to aid artists project turnaround- custom maya scene dispatching scripts using PyMEL- this role gradually transitioned to the Moofe role above