EveryDev.ai
Subscribe
Home
Developers

5,086+ AI companies

  • Radar
  • Trending
AI Tools by Topic
  • AI Coding Assistants
  • Agent Frameworks
  • MCP Servers
  • AI Prompt Tools
  • Vibe Coding Tools
  • AI Design Tools
  • AI Database Tools
  • AI Website Builders
  • AI Testing Tools
  • LLM Evaluations
Follow Us
  • X / Twitter
  • LinkedIn
  • Reddit
  • Discord
  • Threads
  • Bluesky
  • Mastodon
  • YouTube
  • GitHub
  • Instagram
Get Started
  • Users
  • Rate Tools
  • About
  • Editorial Standards
  • Corrections & Disclosures
  • Community Guidelines
  • Advertise
  • Contact Us
  • Newsletter
  • Submit a Tool
  • Start a Discussion
  • Write A Blog
  • Share A Build
  • Terms of Service
  • Privacy Policy
Explore with AI
  • ChatGPT
  • Gemini
  • Claude
  • Grok
  • Perplexity
Agent Experience
  • llms.txt
Theme
With AI, Everyone is a Dev. EveryDev.ai ©
    1. Home
    2. Developers
    3. JJ Eaton

    JJ Eaton

    JJ Eaton is an independent software engineer, maker and AI engineering mentor who builds pragmatic SaaS, micro-SaaS, local-AI systems and custom software. Through BuildThings, he helps businesses apply AI through workflow automation, custom software and private/on-premise AI.

    Visit Website

    At a Glance

    1Tool Listed
    8Products
    9Tool Views
    9Capabilities
    Discussions
    Perth, Western AustraliaHeadquarters
    Focus Areas
    AI Infrastructure
    Model Management
    Local Inference
    Connect
    Latest News
    Published results for Compute Lanes, reporting 6.3-second replies versus 26.9 seconds and 78% fewer computation tokens on some tasks, with no gain on standard benchmarks.Nov 9, 2026
    Published an explanation of why configured AI agents can outperform a single smarter model by reducing token waste and standardising roles.Nov 6, 2026
    Markets
    • Growing businesses needing practical AI automation
    • Founders and teams needing product-minded software execution
    • Businesses with sensitive data that cannot use public-cloud AI by default
    • Companies with local IT partners that need software and on-premise AI implementation
    • +1 more

    AI Tools by JJ Eaton

    (1)
    View DeepSeek-V4.1 TensorFold Spark
    DeepSeek-V4.1 TensorFold Spark tool icon

    DeepSeek-V4.1 TensorFold Spark

    DeepSeek Serving on DGX Spark

    Local InferenceAI InfrastructureModel Management

    Discussions

    No discussions yet

    Be the first to start a discussion about JJ Eaton

    Latest News

    11/09/2026

    Published results for Compute Lanes, reporting 6.3-second replies versus 26.9 seconds and 78% fewer computation tokens on some tasks, with no gain on standard benchmarks.

    buildthings.co
    11/06/2026

    Published an explanation of why configured AI agents can outperform a single smarter model by reducing token waste and standardising roles.

    buildthings.co
    11/04/2026

    Announced an early Compute Lanes feature allowing a local DeepSeek setup to run code while it is still answering; the post reports 38 of 40 coding tests solved and a 16% overall speed improvement.

    buildthings.co
    10/30/2026

    Compared one 512GB Mac Studio with two 256GB machines and argued that two smaller local-AI systems can provide more compute, parallelism and resilience for similar spending.

    buildthings.co

    Products & Services

    8
    BuildThings

    Perth-based AI consulting and software business offering workflow automation, custom AI software, on-premise/local AI and hardware, AI-readiness workshops, ongoing support and a free website-security assessment.

    DeepSeek-V4.1-TensorFold-Spark

    An open-source recipe for serving DeepSeek-V4.1-Flash on two NVIDIA DGX Spark systems with the TensorFold engine. It includes an EXL3 pack, native CUDA/Zig serving, benchmarks, configuration and an OpenAI-compatible API path.

    LocalRouter
    2026-10-08

    A local AI router, scheduler and voice loop for one GPU box, exposed through an MCP server and a lightweight web voice UI. The public repository describes support for local chat, speech, image and video models, loaded on demand.

    SimplifiedPDF

    A browser-based PDF editor for adding text, filling forms, drawing or placing signatures, reordering pages and downloading files locally without signup, subscription, watermark or server upload.

    Market Position

    Eaton positions his work as a pragmatic, product-minded alternative to large consultancies and generic cloud-AI deployments: direct delivery by a Perth-based engineer, maintainable SaaS architecture, and local/on-premise AI where data custody matters. His open-source TensorFold and LocalRouter work also positions him among independent local-inference and self-hosted-AI builders, competing conceptually with vLLM-based local serving, hosted AI APIs and larger AI-automation consultancies.

    Leadership

    Founders

    J(

    JJ (Jayj) Eaton

    Software engineer and founder based in Perth. His published résumé records management/consulting work at SpeedEFX (2017–2022), a Software Engineering Immersive at General Assembly (2022), software engineering at Pricemates, Upwork software engineering from August 2022, and software engineering at VGW/Global Poker from April 2022 to July 2023. His portfolio says he has 15 years in business, five years shipping software and 30+ projects completed.

    Founding Story

    Eaton describes a transition from business management and consulting into software engineering, with an emphasis on combining business context and maintainable implementation. His stated focus is early-stage SaaS architecture and practical AI systems that move quickly without becoming fragile.

    Business Model

    Revenue Model

    Freelance software engineering and fixed-scope/custom AI consulting and implementation through BuildThings, with optional ongoing support. The free SimplifiedPDF and SimplifiedQR tools accept optional Buy Me a Coffee support; BuildThings says audit/workshop scope and pricing are agreed with each customer and private-AI work is sold as a system of software, hardware and model rather than a mandatory subscription.

    Target Markets

    Industries & Segments
    • Growing businesses needing practical AI automation
    • Founders and teams needing product-minded software execution
    • Businesses with sensitive data that cannot use public-cloud AI by default
    • Companies with local IT partners that need software and on-premise AI implementation
    • Developers and enthusiasts operating local AI models on NVIDIA DGX Spark hardware
    Use Cases
    • Early-stage SaaS and micro-SaaS product development
    • AI workflow automation for repetitive business administration
    • Custom internal tools and customer-facing AI applications
    • Private AI for sensitive legal, medical, accounting and commercial records
    • Local inference and multimodal assistants on owned GPU hardware
    • Browser-based PDF form filling and signing

    Quick Facts

    Headquarters
    Perth, Western Australia, Australia

    History & Milestones

    October 26, 2026

    Published a public account of a DeepSeek-V4.1-Flash-1017 build tuned to run on two NVIDIA DGX Spark machines; the recipe and weights were published on GitHub and Hugging Face.

    October 8, 2026

    LocalRouter made its initial public release on GitHub as a local AI routing and multimodal assistant project.

    August 2022

    Began freelance software-engineering work through Upwork.

    February 2022

    Began software-engineering work with Pricemates, a price-alert and historical-price-data product.

    January–March 2022

    Completed General Assembly's 13-week Software Engineering Immersive.

    Key Capabilities

    9
    Full-stack SaaS and micro-SaaS architecture using TypeScript, React/Next.js, Node.js, PostgreSQL and Supabase
    Private/local AI deployment on customer-owned hardware
    AI workflow automation for enquiries, documents, reporting and system hand-offs
    OpenAI-compatible local inference serving and TensorFold-based CUDA/Zig serving
    Local multimodal routing across chat, speech, image and video models
    Client-side privacy: SimplifiedPDF and SimplifiedQR process user content in the browser

    Integrations & Partnerships

    Platform Integrations

    • GitHub repositories and Apache-2.0 licensing for the DeepSeek recipe and LocalRouter
    • NVIDIA DGX Spark hardware with CUDA/TensorFold native serving
    • OpenAI-compatible API serving for the DeepSeek-V4.1-TensorFold-Spark recipe
    • MCP server and browser web UI for LocalRouter
    • Next.js, PostgreSQL, Supabase and Stripe in the Pricemates project
    • Browser-based operation on modern Chrome, Firefox, Safari and Edge for SimplifiedPDF and SimplifiedQR
    • YouTube, LinkedIn, GitHub, X and Upwork as public distribution/contact channels

    Key Partnerships

    TensorFold: the DeepSeek recipe is built on the TensorFold engine, and the project documentation says parts of the recipe have been merged upstream.
    BuildThings says it works alongside Perth IT companies, in-house IT teams and local hardware suppliers rather than replacing the customer's IT partner.
    The DeepSeek repository credits upstream TensorFold contributors and collaborators including Ash Hart and other open-source contributors.

    Connect

    Website
    jayleaton.com/
    GitHub
    jayleaton
    X / Twitter
    jayleaton

    AI Topics

    3

    JJ Eaton focuses on these topics:

    AI Infrastructure(1)
    Model Management(1)
    Local Inference(1)
    Back to all developersSuggest an edit