Pre-Alpha Stage
DeML OS
The infrastructure to execute, verify, settle, and constrain ML autonomously.
Research Scope
Defining the boundaries of decentralized machine learning systems
In Scope
- ML execution under non-centralized scheduling
- Trust and correctness of ML in untrusted environments
- Native pricing and settlement mechanisms that support long-term operation
Out of Scope
- ✗ Systems that only change payment methods without changing ML execution
- ✗ Pure CPU / GPU compute marketplaces
- ✗ AI products or agent applications
System Dimensions
Key evaluation criteria for decentralized ML systems
Execution
Inference, training, and task scheduling mechanisms
Verification
Trust models, correctness proofs, and dispute resolution
Incentive
Pricing models, incentive mechanisms, and sustainability
Governance
Upgradeability, decision-making, and tokenomics
Recent Research Notes
Latest developments in decentralized ML
论文
Mosaic Learning: A Framework for Decentralized Learning with Model Fragmentation
Sayan Biswas Fragmentation
论文
Privacy-Preserving LLM Inference in Practice: A Comparative Survey of Techniques, Trade-Offs, and Deployability
Davide Andreoletti Confidential Inference
论文
Mugi: Value Level Parallelism For Efficient LLMs
Daniel Price VLP
论文
InfiCoEvalChain: A Blockchain-Based Decentralized Framework for Collaborative LLM Evaluation
Yifan Yang LLM Evaluation
论文
Distributed Hybrid Parallelism for Large Language Models: Comparative Study and System Design Guide
Hossam Amer Parallelism
Join the Research Community
If you are interested in distributed systems, ML systems, verifiable computation, or crypto-economic mechanisms at the system layer, you are welcome to join the discussion.