Research infrastructure for distributed intelligence.
DDLSim-Lab is an open-source research environment focused on distributed deep learning, infrastructure behavior, networking, scheduling and fault tolerance.
Project leadership
DDLSim-Lab is organized around an open research model where infrastructure, experiments and implementation can evolve together.
Kaitlyn Brishae Truby
Project Lead & Lead Researcher
Kaitlyn Brishae Truby is the project lead and lead researcher behind DDLSim-Lab, guiding the project's research direction and development.
DDLSim-Lab focuses on the intersection of distributed systems, artificial intelligence infrastructure, networking, scheduling and fault-tolerant computation.
The project is intended to provide researchers and developers with an environment where system-level questions can be explored through reproducible experiments.
Why DDLSim-Lab exists.
Distributed AI systems are influenced by much more than the model itself. DDLSim-Lab brings those system factors into the research environment.
Reproducible experimentation
Create experiments that can be configured, repeated, measured and compared across different system conditions.
Infrastructure-aware research
Treat compute, networking and infrastructure behavior as important parts of distributed AI experiments.
Open research ecosystem
Make the project accessible to researchers, students and developers interested in extending the platform.
Call for infrastructure partners.
DDLSim-Lab welcomes infrastructure contributions that can enable broader and more realistic distributed-system experiments.
Bare-Metal Servers
Compute nodes for systems-scale experiments.
Cloud Credits
Resources for scalable distributed experiments.
Networking Hardware
High-performance networking equipment and testbed access.
Ways to contribute.
Research & infrastructure
Collaboration can take several forms, depending on the available resources and research objectives.
- Provide bare-metal compute resources for distributed experiments.
- Provide access to cloud infrastructure or research computing credits.
- Contribute networking equipment, testbed access or specialized hardware.
- Develop new experiments, workloads, scheduling strategies or analysis tools.
- Improve documentation, reproducibility tooling and open-source infrastructure.
Open by design.
The project is structured so that experiments and infrastructure can evolve independently while remaining part of a common research environment.
Interested in contributing?
Researchers, infrastructure providers, students and developers interested in DDLSim-Lab can get in touch to discuss collaboration and project contributions.