Engineering Manager, Platform Engineering, Deloitte · Detroit metro area & remote
I run a 33-person Platform Engineering org at Deloitte and build AI infrastructure hands-on, from multi-tenant Kubernetes platforms to LLM inference systems.
I build and lead platform engineering organizations. Today I run a 33-person Platform Engineering org at Deloitte, 4 squads with dedicated tech leads and product owners, where I treat the internal developer platform as a product serving thousands of engineers across the US Member Firm.
Outside work I build agent infrastructure hands-on, in TypeScript and Python: multi-agent systems behind approval gates and cost controls, LLM inference serving and performance engineering down to a custom CUDA kernel, and a published design system. The problems I find most interesting are the ones that decide what an agent is allowed to do on your behalf — permissioning, guardrails, and spend — and where those decisions get enforced.
I'm targeting engineering and AI-infrastructure roles — leadership or senior individual contributor — remote, or the Detroit metro area.
2021 – Present
2018 – 2020
2016 – 2018
AI Infrastructure
An enterprise agentic system that triages an IT/Ops queue end to end: retrieves context, drafts grounded responses, and takes guarded actions behind human-approval gates, with RBAC, a hash-chained audit trail, and every release gated by an evaluation harness.
AI Infrastructure
An agentic platform that runs a job search: a dozen specialized agents behind a single autonomy gate that routes every side-effectful action to execute-now or a human approval inbox, with versioned agent memory, cross-provider model fallback, and per-day token budgets enforced before spend.
AI Infrastructure
A platform for LLM inference serving: GPU scheduling, high-throughput serving with vLLM and KServe, and inference-aware autoscaling with KEDA on Kubernetes, with observability and cost/SLO discipline built in.
AI Infrastructure
Inference performance engineering end to end: a device-aware benchmark harness (p50/p95 latency, tokens/sec, peak memory), measured KV-cache, batching, and int8-quantization optimizations, tensor-parallel serving, and a custom fused-RMSNorm CUDA kernel.
An AI orchestrator that simulates a company hierarchy of LLM agents (CTO, VP Eng, engineers) that communicate, produce artifacts, and escalate decisions to the user.
A published React + Tailwind v4 component library (@kornorg/design-system): neutral zinc surfaces, indigo accent, light and dark mode, distributed as an installable npm package.
A full meal-planning and grocery SaaS: recipe discovery, weekly planner, auto-generated grocery lists, grocery-store integration, and subscription billing.
A coordination platform for construction: one shared source of truth for the GC, owner, and subcontractors, with an AI layer that flags cross-trade conflicts before they become rework.
A trip packing-list app: build lists by category, share a trip with travel companions, reuse favorite templates, and let an AI assistant do the busywork.
Also built: graham-crackers (Graham-style value-investing analysis tool), audiobook-generator (local TTS EPUB-to-M4B pipeline), macos-media-scanner (Photos-library cleanup CLI).
I'm open to conversations about senior engineering leadership, platform engineering, cloud, and AI infrastructure roles. The fastest way to reach me is email.