The challenge

Scientific computing capacity is expensive and often concentrated among well-funded organizations, while useful computing resources elsewhere remain idle.

The work

Parallex explores a community-driven framework that connects scientific workloads with idle compute. Its incentive model is designed to make computational research more accessible while using existing hardware more effectively.

Why it matters

Broader access to affordable compute can allow more researchers and small teams to test ideas, reproduce results, and participate in computational science.

The project reported costs up to four times lower than comparable AWS services. Follow the project link for the published project context.

Timeline

Completed milestones are highlighted; muted entries show the work still ahead.

  1. 2023–2024
    Complete

    Peer-to-peer compute network developed

    A five-person Penn Engineering capstone team designed and built a global network for connecting scientific workloads with idle compute.

    Read the Penn Engineering feature (opens in a new tab)
  2. 2024
    Complete

    Leadership Award received

    Penn Engineering recognized the team for its presentation, professionalism, and analysis of the project's commercial potential.

    Read the award coverage (opens in a new tab)
  3. 2024
    Complete

    Cost model validated

    The project reported compute costs up to four times lower than comparable AWS services in its published project context.

  4. Next phase
    Future work

    Broaden real-world pilots

    A future phase could test the network with more compute providers, scientific workloads, and sustained operating conditions.

  5. Future direction
    Future work

    Strengthen scheduling, trust, and incentives

    Further work can harden workload scheduling, provider verification, and incentives for a larger community-run network.

People involved

Project leads, contributors, and collaborators named in the public project record.

  • Vikram Bala

    Project team

  • Ethan Chee

    Project team

  • Anirudh Cowlagi

    Project team

  • Andy Liu (opens in a new tab)

    Project team · AI For Tomorrow Co-Founder

  • Christian Sun

    Project team

  • Boon Thau Loo

    Faculty advisor