Francis – AWS, Kubernetes, Docker

A free web platform providing users with access to open-source book content. The project focused on building an accessible and user-friendly solution for browsing and reading publicly available books online.

• Built and maintained the web platform for accessing open-source book content;
• Developed and maintained the search functionality;
• Cleaned and prepared data for integration into the platform;
• Designed and refined the UI/UX to improve usability and accessibility;
• Developed and integrated new features based on evolving project requirements.

An AI-driven cybersecurity platform designed to ingest and analyze billions of security events across endpoints, cloud workloads, and user identities. The solution supports large-scale security monitoring and threat detection across diverse IT environments.

• Standardized ACLs and managed broker upgrades on Strimzi-managed Kafka clusters running on EKS and GKE as part of the Certified Components program;
• Managed the ZooKeeper-to-KRaft migration by provisioning parallel clusters with MirrorMaker 2 synchronization and gradually repointing clients with offset validation;
• Developed a pre-check script to validate replication lag, client connection counts, and memory pressure before Redis maintenance windows;
• Collaborated with the Kubernetes team to fix anti-affinity rules and ensure Redis primaries and replicas were not scheduled on the same node;
• Built CI/CD gates in GitHub Actions for Docker images and Helm releases, including Terraform plan and policy checks, canary validation, automatic rollback, and post-deployment smoke tests;
• Configured the Vault database secrets engine for Kafka and ClickHouse connections in GovCloud clusters, replacing static credentials with short-lived, rotating tokens;
• Rebuilt monitoring and alerting for shared Redis and Kafka clusters using composite monitors in Datadog, along with existing Prometheus and Grafana signals;
• Strengthened Cassandra access controls with namespace-level network policies and Kyverno rules;
• Investigated a Kafka rebalance issue during rolling broker upgrades and added an upgrade pre-flight check to validate consumer group stability.

An AI-powered regulatory platform that automates FDA 510(k) submission research and preparation. The solution analyzed device descriptions to determine regulatory pathways, identify predicate devices, assess equivalence, hazards, cybersecurity, labeling, and evidence requirements, and generate cited regulatory dossiers and submission forms. The platform included a PostgreSQL rule engine, FastAPI backend, machine-learning services for product-code classification and predicate matching, and a Next.js frontend.

• Designed and built the PostgreSQL rule engine encoding FDA regulatory logic for pathway, predicate, equivalence, hazard, labeling, and evidence assessment;
• Developed the FastAPI backend and integrated ML services for product-code classification and predicate-device matching;
• Built the natural-language Q&A layer, providing evidence-backed answers with regulatory sources;
• Developed the Next.js frontend, including device intake, dossier records, predicate comparisons, and citation-linked views;
• Built an LLM-assisted data extraction pipeline with source verification for high-stakes regulatory information;
• Diagnosed production issues, implemented regression tests, and validated fixes against real user queries;
• Conducted structured reviews using real regulatory cases to identify and improve product accuracy and usability.

A financial services platform for an insurance and financial services brokerage, providing solutions to support insurance management, financial planning, and related client services.

• Built the GRC program in Archer by assigning owners and evidence due dates to control records and organizing them by product line;
• Deployed Okta SSO with Duo MFA for the agent-facing CRM and mapped Active Directory groups to role-based access tiers;
• Wrote a PowerShell script to synchronize broker additions and removals from Active Directory into Okta automatically;
• Led vulnerability management migration from Nessus to Qualys VMDR and tuned scan policies for production-safe coverage;
• Wrote a Python wrapper that enriched findings with business impact and asset owner data and opened ServiceNow tickets automatically;
• Coordinated post-patch validation scans with Kali and Metasploit to support quarterly penetration testing requirements;
• Wrote SPL correlation searches in Splunk ES to detect unusual payout spikes grouped by broker number;
• Exported a feed into Microsoft Sentinel for cloud-side alerting;
• Managed the SOC 2 Type II evidence lifecycle in SecureFrame and mapped artifacts to the Trust Services Criteria;
• Ran recurring secure coding sessions using Burp Suite to walk developers through live OWASP Top Ten flaws on the enrollment portal.