Protocol integrity
Model registries, verifiable compute receipts, access policy and an assurance trail that can survive independent review.
We are assembling a compact senior team around ErisAI and the Ztakora Grid. Six openings cover the contract, runtime, product and hardware boundaries that must work together.
Founded January 2026
AI-native blockchain
On-chain AI runtime
Remote with Singapore overlap
01 / The work in front of us
Model registries, verifiable compute receipts, access policy and an assurance trail that can survive independent review.
Dataset, model and operator workflows connected to provenance, job state, evidence and recoverable compute runs.
A traceable path from protocol and partner requirements to implementation, acceptance evidence and release boundaries.
ErisAI as a runtime-first environment for on-chain inference, receipts, evaluation and selectively portable compute.
02 / Open role directory
The directory is intentionally narrow: two smart-contract roles, one frontend role, one backend role and two cross-product roles.
Protocol engineering
Own the contract boundary for model registries, compute receipts, storage proofs, access policy and governed protocol change.
Protocol security
Own adversarial review, release assurance, privileged-control analysis and incident readiness across the protocol surface.
Frontend engineering
Build dataset, model and compute interfaces that expose provenance and job state without making the wallet the product.
Backend engineering
Build the service layer for dataset metadata, compute-job orchestration, event indexing, receipts and operational reporting.
AI infrastructure
Build ErisAI as a credible on-chain AI runtime first: training, evaluation and inference jobs that produce verifiable receipts.
Product direction
Turn protocol, AI-runtime and partner needs into bounded releases with measurable acceptance criteria and explicit ownership.
03 / Working standard
The work crosses verifiable compute, smart contracts, AI runtimes and hardware operations. We need people who can make boundaries explicit and leave behind useful evidence.
State what belongs on-chain, off-chain and under human authority.
Include blocked, expired, paused, inconsistent and recoverable states.
Architecture and product tradeoffs must be reviewable asynchronously.
Current evidence, design targets and future intent stay visibly separate.
04 / Joining model
Your strongest relevant system, the decisions you owned and the evidence you can share.
A focused conversation about architecture, failure modes and tradeoffs from your prior work.
A live review of one Ztakora problem with the people who would share the boundary.
One to two weeks of genuine project work with an agreed outcome and review standard.
A direct decision on longer-term scope, engagement model and operating expectations.