Michael Marcus Email & Phone Number

- Merging disparate tools into Azure DevOps and Application InsightsIntegrated desperate repos, pipelines, and NuGet feeds into Azure DevOps. Replaced self-hosted agent VMs with MS hosted agents using managed instances. Replaced manual provisioning and maintenance of infrastructure with infrastructure as code by using ARM templates in Azure DevOps deployment pipelines. Integrated separate logging and monitoring solutions into centralised logging and monitoring in Application Insights and Azure monitor.- Static code analysis + security scansIntroduced pull request triggered build validation pipelines that run static code analysis (SonarQube) and fail builds if a quality gate is not met. Created a portfolio structure in SonarQube using regex to automatically move projects into the correct portfolio so that services can be monitored on individual repo, component, application, and product levels. Also added tasks to perform security scans (BlackDuck, Checkmarx) on the pull request triggered pipelines.- Migration of legacy platformCreated a new cluster to replace the outdated Service Fabric cluster used in environments. The new cluster requires a major version match of the OS between Windows Docker containers and nodes rather than an exact patch match, allowing images to be hardened. All environments (dev, test, prod) use the same ARM template with variable changes to enforce uniform states. Deployment moved to Azure DevOps and used managed tasks rather than scripts and self-built console apps making deployments ten times faster.

- Centralised loggingPreviously logs were only generated locally on the VM. I added Application Insights to the full application stack with another developer creating a plugin in the managed CRM which added telemetry from there. This allowed us to trace every operation from page view to potential exception and identify reasons for poor performance. It also allowed us to have automated alerts for certain issues and for testers to pass greater information to developers when tests failed.- Replacing self-hosted infrastructureNew infrastructure could take weeks and the process was completely manual. I replaced this approach by using Infrastructure as code (ARM templates and PowerShell) deploying resources to Azure which allowed environments to be spun up and torn down in minutes, as required. This also removed the need for VM maintenance and made solutions scalable- Training staff to use Azure and Azure DevOpsTeaching developers who have never used Azure or Azure DevOps how to use them to build and deploy applications through multiple environments.- Replacing bespoke solutions with light weight managed alternativesThe previous CI/CD solution was built using C# console apps which required a large maintenance overhead as they needed to be rewritten if testing tools or application components changed. I replaced these with standard Azure DevOps tools and, when not possible, short versioned PowerShell scripts.

London, United Kingdom

In addition to previous responsibilities, as lead I also:- Managed a small teamResponsible for training, planning and prioritising the work of two DevOps Engineers in the Data and Search development team. Providing full lifecycle coverage of the new data platform, two legacy data platforms and several other apps.- Migration of on-prem legacy systems to AzureRather than following a “lift and shift” approach we rebuilt them from scratch leveraging cloud technology for lower costs and lower ongoing support needs i.e. app services instead of virtual machines. This followed the strategy of moving towards SaaS. We also improved components by replacing depreciated technologies and updating underlying components such as Tomcat (and Java) to the latest versions. This increased availability and performance and saved £129K p/a in hosting and licencing costs.- Centralised logging and automating monitoringReplaced the previous centralised logging and monitoring solution (SCM) with Applications Insights. This required resolving the bug we found with one of the app insights SDKs. We resolved this by using the same SDK (portal) across all services and then automating the set-up with an arm template deployed by Azure DevOps- Created live dashboardsBy creating dashboards running from App Insights queries I could monitor the platform at a glance. In the event of issues this provided a good starting point for writing Kusto queries to see what's going on in more detail. It also provided an easy visualisation during a release to make sure everything is still going smoothly.

London, United Kingdom

- Liaising with external developers, data scientists, data journalists and academics, particularly when using our machine readable endpoints- Liaising with the architects, data scientists, data analysts and domain experts who created the data models we use- Migration/transformation of data between data platforms and the APIs of 3rd party data sources- 2nd and 3rd line application and infrastructure support including troubleshooting the root causes and coming up with fixes along with the developers and automating workarounds until the fixes are released- Automating common fixes to problems. Improved existing solutions such as overnight restarts by running them in response to automated availability tests. Downtime came in minutes when it happened and for the first time the system in question achieved over 99.9% availability- Creating CI/CD and Infrastructure as Code. With a goal of instituting CI/CD using Azure DevOps we created the pipelines to build and deploy our software. By using ARM templates we could provision and teardown our infrastructure. This not only meant clean versions were deployed but removed the need to maintain infrastructure with the latest versions of components. Infrastructure chosen was in keeping with a goal to be as close to SaaS as possible i.e. Logic Apps > App Services > Virtual Machines

London, United Kingdom

Responsibilities- 2nd line support (content correction, troubleshooting, fixes) of applications and databases- Deployment of infrastructure and applications- Monitoring of systems- Testing- Database administration- Documentation- Liaising with key business users- Liaising with the development, infrastructure and network teamsKey Achievements- Faster identification of problems via improved queries. Identifying common problems by finding a pattern with the problems so they could be queried rather than spotted individually (SQL, SPARQL) - Automated monitoring. Automating the availability checks that were manually performed by staff on systems by using free 3rd party tools- Knowledge sharing. Documenting and showing other members of the team how I did the above so that they could apply the same techniques to their work- Resolving communication problems between teams. Helping resolve the deadlock between the application support team and the development team that had resulted in a block of the latest release of the data.parliament system. (Largely a result of a toxic relationship that had developed between these teams previously).

London, United Kingdom