Hi, I'm Mike 👋

AI-Accelerated Product Architect — Freelance @ AI Startup (stealth mode) • May 2024 – Present Principal AI Architect & TPM — Cognitive & Agentic Systems @ NGT • Jul 2017 – Present

Building production AI products end-to-end: from mobile apps (iOS/Android) to cloud infrastructure, agentic AI systems, digital humans, physical AI, and BioTech. 20+ years bridging enterprise technology and real-world AI for Fortune 500 clients.

💫 About Me

🔭 I'm currently working on:

Sovereign AI lab — 2x NVIDIA DGX Spark GB10 (Grace Blackwell) + Jetson Orin Nano Super for on-prem model training, fine-tuning, and edge inference benchmarking
Agentic AI systems — hierarchical architectures (Orchestrator → Builder → Specialist → Guardian), MaaS gateway routing, LangGraph/CrewAI, and the NemoClaw stack (Open Harness + Nemotron 3 Super + OpenShell)
Claude Code + MCP — authoring clean SPEC files per CLAUDE.md best practices, sub-agent orchestration for multi-turn code generation, and custom MCP servers (including a HIPAA Compliance Guardian with 63 CFR citations)
Mobile AI shipping — production apps on App Store and Google Play with on-device CoreML/TFLite inference, tested on iPhone 17 Pro Max (iOS Beta)
Snowflake Cortex AI + Snowpark ML — in-warehouse feature engineering, model training, and batch inference processing 10M+ daily records
NVIDIA Omniverse + OpenUSD digital twins — live PLC/MQTT/Kafka telemetry ingestion, Isaac Sim robotics, and synthetic data generation
Digital Humans, Physical AI, and BioTech — production-grade AI products spanning avatar rendering, embodied intelligence, and life sciences workflows

🤝 I'm open to collaborating on:

MCP Gateways and API Gateways — routing, auth, rate limiting, and observability for agent-to-tool and model-to-service traffic
Hybrid Cloud AI Inferencing — on-prem DGX + cloud burst, latency-aware routing, and cost-optimized workload placement
Fine-tuning and deploying NVIDIA Nemotron models — open-weight Nemotron Nano through Nemotron 3 Super, with NeMo Framework and Guardrails
Autonomous agents running in parallel — multi-agent orchestration, shared memory, conflict resolution, and failure recovery patterns
PyTorch → CoreML conversion on iOS with quantization, pruning, and ANE-optimized compression
TensorFlow Lite → Kotlin on Android with NNAPI/Hexagon delegates and INT8/FP16 compression
RAG-as-a-Service — multi-tenant retrieval pipelines with vector DBs, chunking strategies, and eval harnesses
LLM Wiki — curated, versioned knowledge bases for LLM-powered reasoning and grounded generation
Model-as-a-Service (MaaS) — gateway-fronted model routing with vendor optionality (Anthropic, Google, NVIDIA, OpenAI, open-weight)
Blender 3D editing best practices — asset pipelines, USD interop, and Omniverse round-tripping
Advanced Claude system design and solutions — SPEC-driven development, sub-agent orchestration, and MCP-powered workflows

📚 I'm currently learning:

NVIDIA Multi-Modal and AI Networking certifications (full-stack NVIDIA coverage)
MS in AI & Machine Learning at WGU (starts Aug 2026)
Completing BS in Cloud Computing at WGU (Jun 2026)
VisionOS spatial computing and Apple Neural Engine optimization

💬 Ask me about:

Snowflake Cortex AI, Snowpark Python SDK, Dynamic Tables, Streams/Tasks
NVIDIA DGX deployment, NCCL optimization, GPUDirect RDMA, TensorRT
Databricks Unity Catalog, Mosaic AI, Delta Lake medallion architectures
Agentic DevOps with Claude Code, MCP servers, and sub-agent workflows
SCADA/OT data architecture and zero-trust OT/IT segmentation (energy sector)
Full-stack AI apps — Next.js, FastAPI, Flutter/Swift/Kotlin, Firebase, Vercel

⚡️ Fun Fact:

I'm a proud Dog Dad to Kube – yes, he's named after Kubernetes! He occasionally tries to help debug my YAML files to be rewarded with more dog treats. 🐶

🎓 Certifications

NVIDIA (5x): OpenUSD • Agentic AI Professional • GenAI LLM • AI Operations • DGX Administration Snowflake (4x): Architect • SnowPark (Python SDK) • Platform Core • Core (SME contributor, exam item writing 2SOL-C01) AWS/GCP/Azure Cloud: Professional Developer • Cloud Architect • Data Engineer • ML Engineer • Networking • DevOps Enterprise Architecture: TOGAF 9 #194274 (Lifetime) • Databricks ML Professional • AWS Architect • Azure Architect Networking: Dual CCIE #57164 (Enterprise Infrastructure, Service Provider) — August Emeritus 2027 Kubernetes & Linux: CKA • CKS • 3x LPIC-3 #458912 Additional: OpenEDG C++ & Python (Lifetime) • Hyperledger Blockchain • 3M Fiber Optic Journey Man

🧪 Hardware Labs

Personal Sovereign AI Lab

2x NVIDIA DGX Spark GB10 (Grace Blackwell) — sovereign on-prem training/inference nodes with NVLink-C2C fabric pairing
NVIDIA Jetson Orin Nano Super — edge CV and DeepStream demos
NVIDIA Brev Cloud — cloud-burst development environment for hybrid workloads
Mac Mini — 24/7 agentic operations runner for always-on MCP servers, scheduled agents, and background automation
Hugging Face + NVIDIA NGC — pipeline for downloading the latest open-weight models and NIM Blueprints into the lab
MacBook Pro M4 Max — 128GB unified memory, 40-core GPU, 546GB/s bandwidth, nano-texture display
Apple Studio Display XDR — high-fidelity 3D scene rendering for Omniverse/USD composition
iPhone 17 Pro Max (iOS Beta) — TestFlight real-device testing and App Store/Google Play shipping

Customer Labs

DGX BasePOD and SuperPOD clusters up to 1,000 Hopper GPUs with Slurm, NCCL optimization, and InfiniBand HDR/NDR fabric
NVIDIA Blackwell RTX 6000 Pro Workstation and Server platforms from Supermicro for next-gen inference and fine-tuning
NVIDIA Omniverse Nucleus — collaborative USD authoring for digital twin scenes at enterprise scale