Andrew Vincent O'Connor

Senior Site Reliability Engineer focused on AI security, regulated ML infrastructure, and multi-cloud platform engineering. 10+ years in software and infrastructure engineering, with recent work securing PHI-processing medical AI systems across AWS and GCP in HIPAA, SOC 2, HITRUST, and FDA quality-controlled environments. Hands-on experience with private model inference, AI governance and coding-agent guardrails, keyless workload identity, CI/CD supply-chain hardening, GPU training infrastructure, petabyte-scale medical image de-identification, and production observability.

• Own secure deployment, reliability, performance, and cloud infrastructure for FDA-approved AI medical devices processing PHI in HIPAA, SOC 2, HITRUST, and FDA quality-controlled environments.
• Architected a petabyte-scale GCP medical image de-identification pipeline using Cloud Dataflow, batched SQL updates, right-sized workers, and Parquet manifests; reduced run costs ~75% from ~$356K to ~$90K while supporting compliant AI training data preparation.
• Productionized fine-tuned vLLM inference on AWS by packaging 100GB+ Docker images, publishing via GitHub Actions, and deploying isolated SageMaker model containers with no public internet exposure for PHI-bearing inference traffic.
• Provisioned secure 256-GPU NVIDIA H100 Slurm training infrastructure on GCP with private subnet isolation, firewall controls, encrypted shared storage, access/job logging, and Terraform-managed infrastructure for production AI research.
• Established enterprise AI governance and coding-agent safety controls across Claude, Claude Code, GitHub Copilot, and Cursor by deploying Claude Apps Gateway, enforcing engineer-level spend limits and access controls, requiring read-only credentials and PR/Atlantis workflows, and blocking secrets, PHI, destructive Terraform, and mutating AWS CLI actions.
• Introduced governed Terraform-based IaC for previously manually managed GCP environments, using EKS-hosted Atlantis and GCP Workload Identity Federation to enforce PR review, least-privilege access, and keyless cross-cloud deployments.
• Strengthened ML/cloud vulnerability and supply-chain security with AWS Inspector/ECR scanning, Snyk, Dependabot, Terraform/IaC scanning, CVE remediation, hardened AMIs/containers/EC2 instances, immutable patching pipelines, and SHA-pinned GitHub Actions.

• Created reusable Terraform modules adopted across 20+ AWS accounts to standardize CloudWatch dashboards, GitHub Actions OIDC federation, AWS Lambda infrastructure, and AWS Verified Access endpoints for the broader engineering organization.
• Led OIDC-based AWS authentication rollout across ~10 GitHub Actions repos, replacing stored AWS keys with short-lived federated credentials, repo/branch-restricted IAM trust policies, least-privilege access, branch protection, CODEOWNERS, required reviews, and separated Terraform plan/apply workflows.
• Implemented zero-trust access for sensitive internal applications using AWS Verified Access and Okta, replacing VPN-based access for Tableau and Terraform drift detection tools used by approximately 20–30 users.
• Built centralized observability, alerting, uptime dashboards, and incident runbooks for ~30 production systems across 3–4 teams using CloudWatch, Okta-secured access, and PagerDuty, improving incident visibility and MTTR.
• Supported security investigations by helping the security team locate and analyze relevant CloudWatch Insights, CloudTrail, AWS Security Hub, and production telemetry logs.
• Automated AWS operational workflows with Step Functions, Lambda, and Systems Manager for EC2 patching and production operations; extended Terraform CI/CD across AWS/GCP, modernized the EKS delivery platform to reduce cost by ~83%, and built immutable image pipelines for reliable releases and patch management.

• Installed and configured the PerformanceBridge radiology analytics platform and supporting systems across Linux and Windows Server environments for healthcare customers in multiple regions.
• Resolved complex customer concerns and technical issues through deep investigation, research, and reproduction.
• Automated recurring configuration tasks, created procedural documentation, and provided training to colleagues, improving operational consistency and onboarding.
• Supported product expansion into EMEA and APAC through client implementations and technical configuration for organizations including:
  • medneo GmbH
  • King Faisal Specialist Hospital Saudi Arabia
  • Chiba University Hospital

• Built high-performance cloud-based applications for biobanking and laboratory automation.
• Won employee Key Strategy award for contributions to COVID-19 projects.
• Streamlined COVID-19 testing and reporting workflows, increasing efficiency and reducing errors for global biobanking processes.
• Designed a sample lineage interface for the UK NHS COVID-19 system, enabling the processing of hundreds of thousands of tests per week.
• Developed a sample normalization workflow for Curative Korva Labs, which conducted 20% of all tests in California, and integrated with state and local public health departments for reporting results.

• Built Rails-based software for clients including Roche and the National Institutes of Health.
• Reduced sample turnaround times for multiple clients from weeks to hours.
• Designed integrations with hardware devices and external systems including:
  • Scientific instruments (ASTM E1394)
  • Billing interfaces (HL7)
  • Payment systems (Stripe)
  • EHR/EMR systems (Epic)
  • Insurance preauthorizations

• Designed and built analytics tools for the post-acute care industry.
• Built web scrapers to collect plan data from state health insurance exchanges.
• Rebuilt a Tableau-based product using open-source software and on-demand cloud solutions, lowering costs by 50%.

• Designed and developed a medical sample inventory application for an Android RFID reader.
• Integrated with hardware devices including:
  • Barcode readers
  • Signature pads
  • Robotic freezers

Bachelor of Science in Computer Science

AWS SageMaker, vLLM, private AI inference, no-egress model containers, Claude Apps Gateway, enterprise AI governance, AI coding-agent guardrails, Amazon ECR, AWS Bedrock Data Automation, Slurm, NVIDIA H100, GPU training infrastructure, medical imaging foundation models, de-identification pipelines

AWS PrivateLink, VPC endpoints, private subnets, security groups, IAM least privilege, KMS, AWS Verified Access, Okta, GCP Workload Identity Federation, cross-cloud role mappings, workload identity, access controls

AWS Inspector, ECR image scanning, Snyk, Trivy, Dependabot, Terraform/IaC scanning, CVE remediation, hardened Linux AMIs, hardened containers, hardened EC2 instances, immutable image pipelines, SHA-pinned GitHub Actions

Terraform, Atlantis, CloudFormation, Ansible, GitHub Actions, CodeBuild, EC2 Image Builder, CircleCI, OIDC federation, GCP Workload Identity Federation, branch protection, CODEOWNERS, required reviews, governed multi-cloud infrastructure delivery

CloudWatch dashboards, CloudWatch Insights, CloudTrail, AWS Security Hub, PagerDuty, monitoring, alerting, uptime dashboards, incident response runbooks, production telemetry, reliability engineering

Python, Ruby, PostgreSQL, Nginx, GCP Dataflow, GCS, AWS Lambda, Step Functions, Systems Manager, Docker, Linux

HIPAA-regulated systems, SOC 2, HITRUST, FDA quality controls, PHI, medical imaging data, radiology workflows, clinical document processing, HL7, Epic, laboratory workflows

credly.com

drumroll.world
Explore drumming history visually