Rust Engineer — EVM Mempool and On-chain Intelligence Infrastructure

Requirements

• 4+ years in production Rust (preferred) or Go, with low-level systems experience
• Deep working knowledge of EVM internals: calldata, ABI decoding, proxies, traces, ERC-4337.
• Hands-on with major DeFi primitives: Uniswap V2/V3/V4, Curve, Aave, Compound, Maker, Lido, bridges.
• Experience with mempool data at scale: multi-source WebSocket ingestion, propagation asymmetry, replacement/nonce races.
• Track record with event-sourced, append-only systems and deterministic stream processing (Kafka/Redpanda, ClickHouse).
• Solidity at reading level — comfortable decoding contracts without writing them as primary craft.
• Comfort with adversarial and incomplete data — you have debugged a real reorg, not just read about one.

Responsibilities

• Own the Rust data plane: multi-source mempool ingestion, normalization, lifecycle reconciliation.
• Build and maintain the versioned ABI/protocol registry covering major DeFi across supported chains.
• Design the bitemporal entity and transaction graph with append-only assertion-based clustering.
• Implement deterministic, replayable reducers so backfill and reprocessing produce identical state.
• Build coverage profiling, synthetic probes, and the CI determinism canary that audits correctness continuously.
• Define and version the event-schema contract with the Node.js engineer owning the output/analytics plane.
• Extract behavioral signals: wallet clustering, sybil structures, coordination patterns, exploit detection.
   

Project description

Real-time mempool intelligence and behavioral analytics system across multiple EVM chains and L2 rollups. Focus is intelligence-first and correctness-first, not MEV execution — we reconstruct transaction lifecycle and entity relationships under adversarial, partial, and conflicting data sources. The architecture is a five-tier event-sourced pipeline (raw observation log → normalized event stream → lifecycle reconciliation → versioned semantic decoding → bitemporal entity graph) with append-only contracts between every layer. You own the Rust data plane that runs from raw mempool observation to reconstructed entity graph, working alongside a Node.js engineer who owns the output and analytics consumption layer.