Breaking the Bottleneck: A Decentralised Grid for On-Demand Research Compute - Team Dvloper.io at ETHDam
“What if your research task was urgent and horizontally scalable?” “What if you’re new and don’t want to invest in hardware yet?” “What if you could help others reach their computational goals—and get rewarded for it?”
These three questions sparked our weekend sprint at ETHDam. The result is ChainLabGrid: an open source ROFL app that lets anyone contribute or consume compute power, while Trusted Execution Environments (TEEs) keep every contribution secure and every result verifiable.
Framing the problem
Research deadlines rarely wait for procurement cycles. If your model suddenly needs 10× more horsepower, you either overpay a cloud provider or scramble for hardware you’ll soon under-utilise. On the other side of the fence, countless CPUs and GPUs sit idle on laptops, workstations and edge boxes. ChainLabGrid bridges those two worlds: it turns spare capacity into a fluid, on-demand pool while guaranteeing data privacy and result integrity via Trusted Execution Environments (TEEs).
What we actually built at ETHDam
During the 48-hour sprint we delivered a running MVP that was awarded a place in the top 10 applications of the hackathon that:
- Accepts any compute job from a front-end App and records it on a Main Contract.
- Explodes the job into bite-sized subtasks—each gets its own Sub-contract so hundreds of workers can chip away in parallel.
- Lets contributors discover and claim work with a single wallet click; no prior staking or hardware disclosure required.
- Runs every validation step inside TEEs, producing cryptographic attestations that the correct code path executed on untampered data.
- Aggregates and encrypts the final artefact, then releases automatic payments to all successful workers.
A tour of the flow
- Task creation – A user submits a compute request; the Main Contract stores metadata and notifies an off-chain Request Listener.
- Task expansion – The listener verifies the payload, updates the active-task ledger and calls back to mint individual Sub-contracts.
- Discovery & assignment – Contributors pull a live list of open work, pick a subtask, and the contract assigns it atomically.
- Execution – The contributor’s node crunches numbers; capacity checks prevent over-commitment.
- Validation – Sub-Task Validators inside TEEs replay the job deterministically; only approved outputs move forward.
- Aggregation & payout – Once every piece passes, a Task Aggregator stitches results, encrypts them for the requester, and the chain disburses rewards.
Why TEEs instead of classic crypto-proofs?
- Privacy first – sensitive research data never leaves enclave memory.
- Lightweight trust – we sidestep heavy zero-knowledge proofs; the hardware attestation is both cheaper and faster to verify.
- Interoperability – by building on Oasis Sapphire (EVM-compatible), we reuse the Solidity ecosystem while inheriting enclave guarantees.
The impact we’re chasing
ChainLabGrid turns compute into a liquid commodity: researchers rent milliseconds instead of machines, newcomers monetise idle rigs with two clicks, and enterprises gain a privacy-preserving overflow buffer for bursty workloads. All secured by the same enclave tech that protects mission-critical fintech and healthcare systems.
