Skip to content
James Foss, jamesfoss.me home

Full-Stack Developer & Systems Engineer

JamesFoss

Building intelligent applications, scalable infrastructure, and modern digital experiences.

  • Full-Stack Development
  • AI Systems
  • Infrastructure
  • Automation
Explore My Work
Get In Touch
01About

Beyond Code.
Into Systems.

I'm a full-stack developer and Linux systems administrator. I build modern web applications, run the production infrastructure they live on, and develop self-hosted AI systems on local GPU hardware.

Through Linear Web Solutions LLC, the agency I founded and run, I design, build, and manage digital systems for businesses. The work runs the full length of the stack: interfaces and APIs, servers and networks, virtualization, security, deployment, and the AI and automation layered on top.

Working on both sides of the deploy button means fewer handoffs, fewer blind spots, and systems that are designed as a whole rather than assembled from parts.

Operating principle

I don't just build websites. I engineer complete digital systems.

James Foss
  1. 01Interface
  2. 02Application
  3. 03Services
  4. 04Infrastructure
  5. 05Compute
Fig. 02 — Working range, end to end
  • A.01

    Application Engineering

    Full-stack web applications in TypeScript, React, and Next.js: server-rendered interfaces, typed APIs, and integrations with the services a business depends on.

    • TypeScript
    • Next.js
    • Node.js
    • REST APIs
  • A.02

    Linux & Infrastructure

    Production Linux servers, Proxmox virtualization, networking, and the reverse proxies and process managers that keep applications online.

    • Ubuntu
    • Proxmox
    • Nginx
    • SSH
  • A.03

    AI & Automation

    Self-hosted language models on NVIDIA GPUs, agent workflows, and automation pipelines that turn repetitive work into scheduled jobs.

    • Ollama
    • Langflow
    • GPU
    • Agents
  • A.04

    Deployment & Operations

    Containerized builds, Git-based deploys, health checks, and the operational discipline that makes releases uneventful.

    • Docker
    • Coolify
    • PM2
    • Bash
02Expertise

From the browser
to bare metal.

No percentages and no self-ratings. Capability shows up in what ships: pick a project to see the tools it runs on, or pick a tool to see where it is used.

31 technologies across five disciplines. Select a project or a tool to trace where each one is actually used.

01Languages

7

Typed where it matters, scripted where it's faster.

02Frontend

4

Fast, accessible interfaces that hold up on any screen.

03Backend

4

Services, data, and the integrations between them.

04Infrastructure

9

The machines, networks, and platforms underneath.

05AI Engineering

7

Models served locally and put to work.

03Projects

Selected work.
Built and running.

Live products, private systems, and work in progress, each labeled for what it is. Private systems are described without exposing internals.

  • Live
  • Private system
  • In development
P.02Private system

Operations platform

LWS Dashboard

A centralized application for managing websites, business workflows, content automation, and connected services.

  • Next.js
  • React
  • TypeScript
  • JavaScript
P.04Private system

Local LLM platform

Self-Hosted AI Infrastructure

A locally hosted AI environment incorporating NVIDIA GPUs, Ollama, large language models, and AI workflows.

  • Ollama
  • Local Large Language Models
  • GPU Computing
  • Model Deployment
P.05Private system

Virtualization platform

Proxmox Infrastructure

A virtualization platform supporting Linux servers, production applications, networking, and private services.

  • Proxmox
  • Virtualization
  • Linux
  • Networking
P.06In development

Automation pipelines

AI Content Automation

An automated workflow system for content generation, publishing, and multi-platform distribution.

  • AI Automation
  • TypeScript
  • Node.js
  • REST APIs
Browse the project index
04Infrastructure

The Systems
Behind the Code

A sanitized, conceptual view of the self-hosted stack I design and operate, from a request arriving at the edge to inference on a GPU. Select a layer or component to see what runs there, or trace a request end to end.

RequestResponse

    Browsers, phones, and API consumers. Interfaces are server-rendered where possible and hydrated only where interaction needs it, so the first response is already useful.

    Select a component for detail.

    Technologies

    • Next.js
    • React
    • TypeScript

    Seen in

Client Applications.

Conceptual and sanitized. No hostnames, addresses, ports, credentials, or configuration from production systems appear on this site.

05AI Systems

Intelligence.
Engineered Locally.

Language models running on my own NVIDIA hardware, wired into agents and automation that do real work. Private by default, and built to be operated rather than demoed.

  1. 01 · Input

    Brief

    A topic, draft, or data record enters the pipeline.

  2. 02 · Model

    Local LLM

    A model served by Ollama drafts text on the GPU.

  3. 03 · Processing

    Structure

    Output is validated, formatted, and given metadata.

  4. 04 · Automation

    Schedule

    Jobs queue rendering and publishing for the right moment.

  5. 05 · Output

    Publish

    Posts, cards, and video reach connected platforms.

pipeline.logSimulation · no model connected
Idle

$ ready. Run the simulation to walk a brief through the pipeline.

Illustrative only. This scripted demo runs in your browser; no model, queue, or platform is connected.

  • AI.01

    Self-hosted LLM deployment

    Language models served with Ollama on local hardware and exposed to applications through a private API.

  • AI.02

    GPU-accelerated inference

    NVIDIA GPUs handle inference. Matching model size and quantization to available VRAM is part of the engineering.

  • AI.03

    AI agent architectures

    Multi-step agents and visual workflows in Langflow, composed from models, tools, and data sources.

  • AI.04

    Workflow automation

    Scheduled jobs and dedicated workers that run business processes end to end without manual steps.

  • AI.05

    AI-assisted development

    AI coding tools built into day-to-day development, from scaffolding to refactoring, with engineering judgment in charge.

  • AI.06

    Automated content pipelines

    Pipelines that select, caption, render, and publish cards and short-form video across platforms on a schedule.

06Contact

Let's Build Something Extraordinary.

Have an application to build, infrastructure to untangle, or a process worth automating? Send a few details and I'll reply personally.

Email
james@linearwebsolutions.com
Agency
Linear Web Solutions LLClinearwebsolutions.com(opens in a new tab)
Based in
Southern Oregon, USA
Time zone
Pacific Time

New message

Validated on the server

What's it about?
0/5000

Prefer email? james@linearwebsolutions.com