QuickMDSim

Roadmap

What’s shipping, in order

Goal: signup to a downloaded dump in under 10 minutes, then come back next week without touching a cluster. This list moves when priorities do.

Shipped

  • Cloudflare re-platform

    Shipped

    Workers, D1, R2, Queues, Pages — serverless by default.

  • Accounts & auth

    Shipped

    Email/password signup and login with JWT sessions.

  • Projects, files & jobs UI

    Shipped

    The core project → edit → run → download loop.

  • Free tier + starter CPU-hours

    Shipped

    0.5 CPU-hour seed after email verify; overage $0.25/vCPU-hour.

  • Stripe billing (Basic & Pro)

    Shipped

    Monthly/yearly plans and CPU-hour credits.

  • Production LAMMPS runner

    Shipped

    Every job runs genuine LAMMPS in a Cloudflare container.

  • Job sizes 1 / 2 / 4 cores

    Shipped

    Pick cores per job. Free stays on 1; paid can use 2 or 4 (Cloudflare max is 4 cores / 12 GB). Credits burn per core.

  • Authenticated MCP server

    Shipped

    Personal access tokens + remote MCP. Agents (Hermes, Claude, Cursor) can create projects, write inputs, submit jobs, and read logs.

Building next

  • First-job tutorial

    Planned

    A canonical 'first job in 10 minutes' guide, linked from the marketing site and app empty states.

  • Output previews

    Planned

    Inline preview for common output file types — logs, images, and movies right in the job view.

  • Multi-file projects

    Planned

    Polish for workflows that span data files, force fields, and include files.

  • Job observability

    Planned

    A founder dashboard for job-failure reasons — so we fix the platform before you have to.

Later

  • Restart-file workflows

    Pick up a job from a checkpoint instead of starting over.

  • Course & lab packaging

    Instructor guides and isolated short jobs for teaching MD.

  • Team & org accounts

    Pooled credits and shared projects for labs and small teams.

  • Collaboration

    Read-only project sharing with collaborators.

Not this, not yet

  • National-lab scale (10,000-core) production HPC parity
  • Non-LAMMPS engines (GROMACS, OpenMM) in the near term
  • On-prem / FedRAMP enterprise install
  • A click-and-drag GUI that hides LAMMPS input language

If you need something on this list, tell us.