Rich Gates · AI systems & product engineer

I build reliable AI products, agent systems, and local infrastructure.

I work across model APIs, agent runtimes, multimodal pipelines, product interfaces, data, deployment, and recovery. The goal is not a convincing demo. It is a system a team can understand, operate, and improve.

Review selected workProfessional profile
SYS_001Routing to reliable outcomes
A central system spine connecting models, agents, tools, runtimes, product interfaces, and infrastructure.Agent layerTools & actionsRuntimeModel interfaceLocal inferenceData & contextDeploymentObservability
Architecture noteObservable by design. Determinism where possible. Recovery always.

01 / Professional profile

Systems thinking with a product finish.

I work across the entire path from model to outcome: provider contracts, agent tools, multimodal workflows, local inference, product interfaces, data, deployment, and recovery. I am most useful where several technical layers must become one coherent, reliable product.

What I build
AI products, agent operations, multimodal workflows, and local model infrastructure.
Where I work
Across interfaces, APIs, data, model runtimes, deployment, and recovery.
How I deliver
Grounded in real system state, explicit boundaries, measurable proof, and operable handoff.

02 / Selected work

Serious systems, clearly explained.

Six case studies covering product design, agent operations, multimodal infrastructure, data systems, and local compute.

01
Agent systemsActive development

Xorvio

A cross-platform command surface for models, agents, tools, artifacts, and local compute.

  • Tauri 2
  • Rust
  • React 19
  • TypeScript
  • Vite
View case study
02
MultimodalProduction foundation / active development

PromptSilo

A generative-media product that evolved from prompt marketplace to continuity-aware, provider-orchestrated filmmaking workstation.

  • Next.js App Router
  • React 19
  • TypeScript
  • Vite
  • Express
View case study
03
KnowledgeShipped self-hosted system

FeedSilo

A durable research system that turns noisy public feeds into multilingual, multimodal, searchable knowledge.

  • Next.js
  • React
  • TypeScript
  • Prisma
  • PostgreSQL
View case study
04
Agent systemsActive integration and operations program

Hermes / OpenClaw / Legion

The operational layer around agents: brokers, tools, browser sessions, collaboration protocols, local GPU services, and recovery.

  • Hermes
  • OpenClaw
  • MCP
  • React
  • Node.js
View case study
05
Media intelligenceAdvanced active prototype

Aural Echoes

A deterministic policy brain and broadcast architecture for a 24/7 AI-assisted station.

  • TypeScript
  • SQLite
  • Semantic HTML
  • Custom CSS
  • Browser Audio
View case study
06
Local computeActive integration program

Local AI lab

Making heterogeneous local models behave like dependable product infrastructure.

  • OpenAI-compatible APIs
  • Responses API
  • Chat Completions
  • SSE
  • MLX-family runtimes
View case study
View all case studies

03 / Technical stack

Current tools. Durable fundamentals.

Broad enough to connect the full system; selective enough to use the right tool for the constraint.

01

AI & agents

  • OpenAI and Anthropic protocols
  • MCP, tools, memory, delegation
  • Multimodal generation and analysis
  • Local model compatibility routing
02

Product engineering

  • React 19, Next.js, TypeScript
  • Tauri 2, Rust, SwiftUI
  • Node.js, Python, FastAPI
  • Accessible responsive interfaces
03

Data & platforms

  • PostgreSQL, Prisma, pgvector
  • SQLite, FTS5, D1, Drizzle
  • Queues, SSE, WebSockets
  • R2 and S3-compatible storage
04

Infrastructure

  • Docker, Linux, macOS, Unraid
  • NVIDIA, CUDA, Metal, PyTorch
  • Cloudflare Workers and Sites
  • Playwright, Vitest, deploy verification

04 / Working approach

Built to be understood, operated, and improved.

  1. 01

    Inspect reality first.

    The live process, deployed copy, broker, database row, model artifact, or browser session is the source of truth.

  2. 02

    Build the connective layer.

    The difficult work often lives between model, tool, worker, product, and infrastructure.

  3. 03

    Make failure legible.

    Capability boundaries, retries, partial success, health checks, receipts, and honest unavailable states beat silent guesses.

  4. 04

    Keep people in control.

    Human review gates, rights checks, approval boundaries, privacy-safe state, and reversible operations are product features.

05 / Contact

Open to serious technical work.

If you need someone who can connect AI capability to a dependable product and its operating environment, I would like to hear about the problem.

Professional contact details available on request.