John Pellman Email & Phone Number

New York, Ny, Us

New York, Ny, Us

New York, Ny, Us

New York, Ny, Us

• Migrated end-users from an on-premises Hadoop cluster running Cloudera Distributed Hadoop (CDH) to a Dataproc cluster running in GCP. To ensure a low-friction transition, I created a cloud architecture that mimicked the on-premises cluster as closely as possible. This architecture made use of persistent disks to store HDFS data, Cloud SQL for the Hive metastore, and OS Login for identity management. Additionally, I developed a static site to serve as a portal with human-readable redirect links to various Dataproc component web interfaces (dataproc.hpc.nyu.edu). All cloud resources were codified in Terraform files.• Wrote descriptions for various NIST 800-171 controls as part of an effort to create a cloud-based secure research environment for sensitive data (e.g.,controlled unclassified information, HIPAA limited data sets). Also wrote descriptions for 110 NIST 800-171 controls for an on-premises secure environment based on OpenShift. Contributed to CMMC (Cybersecurity Maturity Model Certification) 2.0 Level 2 certification process.• Implemented two GCP reference architectures using Terraform: one to perform virus scans using ClamAV on items ingested into a bucket, and another to perform data loss prevention (DLP) scans on items ingested into a separate bucket.• Implemented a GCP Cloud Function to automatically turn off virtual machine instances after a budget threshold is crossed. • Implemented an ad-hoc Apache server for password-protected file-sharing.• Supervised and mentored a student worker.

New York, Ny, Us

· Maintained a GPU cluster used by several research groups at the Zuckerman Institute. Debug GPU issues (often involving IOMMU or GPUs disappearing from the PCIe bus) on cluster nodes.· Researched and implemented changes to the cluster's SLURM scheduler configuration to meet researcher expectations. Implemented custom scheduling logic to increase cluster responsiveness. Maintained regular communication with key stakeholders during changes.· Implemented FS-Cache on GPU cluster nodes to increase read speeds from NFS-based storage of machine learning datasets consisting of many small files.· Created custom wrapper scripts to display available cluster resources in a more user-friendly output (see https://github.com/ZuckermanBrain/axon-slurm-customizations).· Configured a separate SLURM instance for an individual lab with integration with the RELION CryoEM software package.· Maintained Ansible roles and playbooks to reliably reproduce configurations (for SLURM, identity management, etc; see https://github.com/search?q=topic%3Aansible-role+org%3AZuckermanBrain+fork%3Atrue).· Put together and taught a course on Git / Version Control adapted from the Software Carpentry curriculum (see https://zuckermanbrain.github.io/git-novice/). Major revisions to the curriculum included discussions of Columbia-specific policies and an emphasis on core concepts rather than command line syntax.· Configured a Guacamole-based remote desktop portal.

New York, Ny, Us

· Managed 140 Linux hosts (bare metal servers, Xen VMs, VMWare VMs) in a heterogeneous computing environment (CentOS 6, CentOS 7, Ubuntu 14.04) at Columbia University Libraries.· Codified infrastructure using Ansible and Saltstack.· Implemented monitoring using the TICK stack and Grafana across libraries hosts.· Scraped Google Groups content for long-term preservation.· Benchmarked an application / data processing pipeline for producing archival packages (Archivematica) across various AWS EC2 instances to find an optimal cloud-based configuration. Codified this configuration in Terraform.· Revised code used to automatically provision and deprovision user accounts.· Deployed an internal-facing Zulip (Slack alternative) chat server for communication between teams.

New York, Ny, Us

· Performed user support for the Configurable Pipeline for the Analysis of Connectomes (C-PAC), a Python-based package for analyzing large-scale rs-fMRI datasets with a user-friendly interface.· Updated the user documentation for C-PAC, various other software packages, and the Rockland Sample, an open access fMRI dataset.· Maintained an Amazon Web Services virtual machine image with C-PAC and all of its dependencies.· Provisioned from scratch two servers. Migrated users over from an older configuration. Documented the steps needed for our organization to provision future servers similarly in an executable shell script.· Organized neuroimaging data into a standardized directory scheme (the Brain Imaging Data Structure) using a Python script. Contributed to the development of the standard through debugging a validator program used on the re-organized data and proposing conventions for organizing data acquired using aparticular MRI sequence.· Troubleshooted GRUB bootloader.· Set up Jenkins and Docker on a server to perform unit and regression tests on new builds of C-PAC.· Configured Google Analytics for C-PAC’s website.· Compared quotes for a storage server replacement and suggested vendors.· Created a simple system based on cron and gpg for encrypted tape backups. Advised on the setup of Bacula as a more long-term solution.· Prepared hard drives to be secure by re-formatting them using LUKS encryption.· Consolidated two physical drives into a single LVM partition on a Linux server.

Postbac Studies, Computer Science

B.A, Psychology; French; Cognitive Science

Junior Year In France (Study Abroad Program)

Education record

Education record