Josef Doornink

What happens when you deploy a non-deterministic reasoning engine in a system that requires guarantees? K8gentS is an autonomous Kubernetes RCA agent built around that question. It routes cluster failures through Gemini-powered analysis, gates remediation behind both a human approval and an OPA Gatekeeper admission policy, and exposes diagnostics via an MCP server published on the official MCP Registry as io.github.JDoornink/k8gents. See the README for an open discussion of failure modes and confidence calibration tradeoffs.

A self-healing multimodal search engine demonstrating the agent-in-the-reliability-loop pattern applied to ML systems. CLIP embeddings + Qdrant power semantic video search ("find a red truck at night"); Prometheus and Evidently AI surface drift; an LLM agent reasons about drift signals and triggers automated retraining via GitHub Actions. Companion to K8gentS — same thesis (LLM as decision-maker inside a reliability contract), different domain.

Static analysis for the agent supply chain. A published Python CLI tool that validates MCP servers and scans AI agent implementations for security vulnerabilities — configurable security levels, CI/CD integration with threshold-based failure conditions, and SARIF output for integration with existing security tooling.

The source code driving this exact platform. A Python pipeline that uses Claude to parse unstructured job descriptions and output statically generated, tailored frontend bundles via Next.js — the dogfood project for the Agentic SRE thesis.

• —Designing and building the AKS-based ML training and inference platform — making the foundational architecture decisions for an early-stage AI company before production traffic.
• —Establishing patterns for distributed training resource management on Azure, balancing cost and iteration speed for research workloads.
• —Translating research-team requirements into cloud architectures, working iteratively as the platform and the model strategy co-evolve.
• —Standing up the distributed training pipeline from scratch — automated pipeline scripts, validation steps, and the reliability scaffolding research teams need to move fast.

• —Implemented New Relic observability stack with distributed tracing, cutting MTTR by 50%.
• —Built automation tooling in Python and Go eliminating 80+ hours/month of toil and accelerating deployment velocity 3x.
• —Architected AKS production environments handling 14K+ requests/day across 30+ microservices at 99.9% uptime SLA.
• —Built and maintained Human Resources Information management system, responsible for over $8.3M+ in ARR and 43% YoY growth rate.
• —Led Kubernetes capacity planning and strategies supporting 200% traffic growth.
• —Built CI/CD pipelines (GitHub Actions, Azure DevOps) with automated testing and rollback mechanisms.
• —Managed 100+ cloud resources via Terraform IaC; implemented CKS security controls for SOC2 compliance.

• —Developed cloud-based SaaS applications using .NET and Angular, migrating on-premise software solutions to Azure cloud platform.
• —Built RESTful APIs for multi-tenant applications serving thousands of users with focus on performance and scalability.

• —Engineered multi-tenant e-commerce platform using Microsoft Stack (.NET, C#, SQL Server) integrated with third-party SaaS APIs.
• —Led 'uplift' initiative migrating legacy codebase to modern greenfield platform, improving response times by 40% measured through New Relic APM.

• —Lead Test and Development Engineer for NIH-funded multimillion-dollar research project focused on bone fixation solutions.
• —Managed successful implant creation, delivery, and test methodology producing multiple US FDA-approved implants.