Skip to content
Quiet Consensus

Selected experience and research

Work that informs how we approach new problems.

Production engineering, infrastructure delivery, protocol research and experimental systems across distributed computing, blockchain and applied AI.

A selection of technical work and research informing the practice.

Professional experience

Production engineering, architecture and technical leadership on institutional infrastructure.

  1. Technical product leadership · architecture and readiness

    Research-to-production leadership for institutional infrastructure

    Technical product leadership for a privacy-oriented, EVM-compatible ledger designed for capital-markets institutions, as it moved from inherited research code towards production.

    Challenge
    Code built for research demonstrations had to become infrastructure that regulated institutions could depend on, against fixed external timelines.
    Contribution
    Co-led with the head of engineering. The work involved owning product requirements, phased release scope, readiness gates with explicit exit criteria, genesis planning, the launch runbook and performance-test design. Treasury and signing architecture, validator topology and RPC exposure controls were specified with the engineers who implemented them.
    Approach
    Explicit trade-offs over aspiration: reliable core operations ahead of new features, a migration to a less mature framework deferred, and performance testing framed around bottlenecks and tail finality rather than headline throughput. Significant decisions went into an architecture decision register.
    Outcome
    Readiness criteria, launch and verification procedures, and architectural decisions supporting the move from testnet towards production.

  2. Hands-on production engineering · Go

    Cross-chain routing API

    Production Go engineering for a cross-chain DeFi routing API, replacing destination-specific integrations with a single OpenAPI plugin contract.

    Challenge
    The API had grown one integration at a time, each with its own interface, and API-gateway configuration was a manual step in every release.
    Contribution
    Engineering work included implementing production Go services and bridge integrations, including Stargate and LayerZero, against a shared OpenAPI plugin contract; moving gateway policies into the specification itself; and automating specification releases through CI/CD. It also included tracing a routing-reliability defect in a bridge integration to its root cause and driving the fix.
    Approach
    The API contract as the source of truth: integration interfaces, gateway behaviour and releases all derive from the specification. Routes were prioritised by chain reliability, latency and usage.
    Outcome
    A manual configuration step removed from every release, a consistent interface for new integrations, and reliability-based route prioritisation that reduced support tickets.

  3. Technical product ownership and analysis

    Validator performance and rewards

    Product ownership and on-chain analysis for institutional staking: rewards benchmarking, block timing and protocol upgrades.

    Challenge
    Where were rewards being lost across a large institutional Ethereum validator fleet, how did it compare with the wider network, and what would protocol upgrades change?
    Contribution
    Product ownership covered the protocol-rewards roadmap across Ethereum, Solana and other networks, block-timing and latency strategy, and institutional guidance on Ethereum’s Pectra upgrade (EIP-7251). Analysis included consensus- and execution-layer rewards benchmarking with Dune, BigQuery and on-chain data, evaluation of MEV relay configuration and validator-consolidation economics, and Solana cohort benchmarking. The validator nodes themselves were operated by a dedicated node-operations team.
    Approach
    Rewards were decomposed by source — attestations, proposals, sync committees, execution fees and MEV — and compared across own-fleet, network and peer cohorts, so each finding pointed to a specific configuration or roadmap decision.
    Outcome
    Analysis that informed block-timing strategy, MEV configuration and the product roadmap, including scoping tooling to replace a largely manual latency-configuration process.

Research

Protocol research, experimental distributed systems and applied AI.

  1. Protocol research

    Layer 2 sequencing and Ethereum protocol research

    Research on Layer 2 block building and ordering, decentralised sequencing, Ethereum client APIs and protocol adoption.

    1. Decentralised sequencing

      Question: What does taking part in shared sequencing involve for an institutional infrastructure provider?

      Investigated an early decentralised-sequencer architecture over roughly a year, examining its operational requirements, trust assumptions and implications for infrastructure providers. The findings were published and informed a testnet deployment.

    2. Proposer-builder separation on Layer 2

      Question: How do competing rollup designs compare on who builds and orders blocks?

      Compared six Ethereum rollups on centralisation, censorship resistance, MEV distribution, fees and inclusion time, producing a scoring rubric for PBS designs and a Go proof of concept of MEV-aware block construction.

    3. Ethereum client APIs

      Question: How consistently do clients cover the Ethereum JSON-RPC API, and could OpenAPI tooling make it easier to build on?

      Research-fellowship tooling that maps 65 JSON-RPC methods across major clients, with an OpenAPI-based approach to the specification.

    4. Protocol adoption

      Question: Why was a technically promising Ethereum scaling approach not being adopted?

      Grant-funded interviews with more than 20 engineering teams (2018–19) identified specification, documentation and tooling gaps as the main barriers.

  2. Experimental distributed systems · Rust

    Parallax: consensus, deterministic execution and verifiable settlement

    A Rust research prototype of a private financial ledger: a deterministic order-book engine under Malachite BFT consensus, with zero-knowledge proofs of execution gating settlement and experimental post-quantum hybrid cryptography.

    Challenge
    Can one network combine BFT finality, a private high-throughput state engine, validity proofs of execution and cryptographic agility?
    Approach
    Each phase was framed as a pass/fail question — determinism, replication, consensus, settlement, proofs — with decisions recorded in architecture decision records and negative results reported as such.
    Outcome
    An unaudited research proof of concept, benchmarked on a single machine. It is not production software.

  3. Internal tooling and experiments

    Applied AI for engineering

    Standards-research automation, AI-assisted investigation and machine-learning experiments on on-chain data and consensus.

    Contribution
    Internal tools: an LLM-based agent, used by product and protocol engineers, that synthesised Ethereum research forums and GitHub activity to rank emerging standards; and AI-assisted analysis to extend technical investigations, with findings verified by engineers. Experiments: Isolation Forest anomaly detection over recent Ethereum transactions, self-hosted agent-based research automation with MEV monitoring and LLM summaries, and ML-assisted BFT leader selection on synthetic data.
    Outcome
    Tools and experiments for testing ideas quickly — not validated production AI systems.

Contact

Working on something in this space?

If your challenge overlaps with this experience, we’d be glad to hear about it.

Get in touch