At U.Group I was part of a Platform Engineering team for a major U.S. Federal Agency. There I contributed to standing up a secure, scalable, resilient multi-TB ElasticSearch cluster on AWS utilizing EC2 and EBS. My team built and maintained multi-environment infrastructure in AWS (EC2, IAM, KMS, Kinesis, DynamoDB, S3, and more) as well as multi-environment infrastructure in OCP. This infrastructure was managed via Terraform. As part of this effort, we built out shared, reusable Terraform libraries and wrote custom Terraform providers in Golang. I built out security-hardened Ubuntu 18 AMIs, as well as application specific AMIs for New Relic and ElasticSearch, using Packer and Chef. All of the work done on this team was automated through Jenkins CI/CD pipelines.

I transitioned off of the PowerVM OpenStack drivers and started a new role with the emerging PowerAI team. My squad (PIE - PowerAI Inferencing Engine) was initially tasked with an effort to run machine learning training and inferencing algorithms on alternative processing endpoints (including GPUs, FPGAs and machine learning specialized GPUs). This involved running containerized (Docker, Kubernetes, Helm) machine learning services. I wrote python pre and post-processing algorithms that allowed the inferencing and training to occur at lower (16 or 8) bit-widths. This allowed us to vastly increase algorithm efficiency, which in turn allowed object detection and classification of video streams in addition to still images (PowerAI Vision). Our squad then transitioned and we were tasked with integrating the IBM Watson frontend with the PowerAI inferencing service. This involved containerizing services and taking research-level code-bases and refactoring them to be production-ready.

After graduating in 2016 I accepted an offer to continue working full time for IBM. I continued to be the project owner for the PowerVM CI pipeline. I dramatically increased the resiliency of the two CI pools (staging and production) and vastly improved our cloud driver code coverage. At this point I became system admin for over 20 POWER and x86 servers, running up to 6 different ubuntu server configurations. The deployment of the servers became fully automated using Ansible, greatly improving environment deployment time. I also became heavily involved in an effort for the PowerVM OpenStack drivers to become officially supported by the OpenStack community. This involved taking IBM driver code and refactoring it to conform to the OpenStack community standards, eventually allowing the driver code to be open source.

Following my 12 week summer internship, I was offered the opportunity to continue working as a part-time IBM employee during my senior year of college. For this role, I picked right up where I left off and continued to work on the PowerVM CI pipeline. Only working 10-20 hours per week, my main responsibilities were ensuring CI resiliency and uptime, as well as fixing or contacting the various code owners regarding bugs the pipeline uncovered.

This was a 12 week software engineering internship between my junior and senior year of college. I performed a variety of testing and bug-fixing tasks for the PowerVC and PowerVM teams, familiarizing myself with my first production code-bases. My main project over the internship was developing the Jenkins based Continuous Integration (CI) pipeline for the PowerVM OpenStack compute, networking, and storage cloud drivers.