Eric Shi / @shikelabs
Building AI-native infrastructure for blockchain developers.
Parallel execution. zkVMs. Distributed runtimes. Protocol architecture.
Current Thesis
The execution layer is the most under-engineered component in modern blockchain architectures. Parallel execution, deterministic state machines, and runtime observability are not optimizations — they are prerequisites for any execution environment that claims to be a credible settlement layer.
Writing
all writing →AI Still Cannot Design Correct Ledger Systems
2026-05-30LLMs can write code. They can flag known vulnerabilities. They cannot reason about adversarial state transitions, emergent multi-contract bugs, or mechanism design failures. The gap is the size of the entire blockchain security incident database.
Why Most Parallel Execution Designs Fail
2026-05-28Parallel execution is a checkbox on every L1 roadmap. The benchmarks look great. Production tells a different story. An honest look at why — and what actually matters.
Notes
all notes →There's a growing observability gap in blockchain infra. Systems are getting more complex but our tools haven't kept up.
Modularity gives you flexibility at the cost of integration complexity.
Nobody talks about recovery semantics until they need them.
Infrastructure economics are weird in crypto. Technically superior infra can lose to worse but established solutions.
Protocols have lifecycles: design -> implementation -> deployment -> ossification -> technical debt -> crisis.
Projects
all projects →finova-ai ↗
activeAI-powered financial analysis and cash flow forecasting for SMBs.
parallel-execution-models
plannedModels and benchmarks for parallel transaction execution in blockchain systems.
zkvm-runtime-experiments
plannedExperiments with zkVM runtime design and deterministic execution.