AI Engineering.
Agent Systems.
Evolving Systems.

Agent integration, AI-assisted dev workflows, and protocol hardening.
Delivered as focused sprints with clear deliverables.

ex-Squidrouter ($500M volume) · Led AI workflow adoption at FastLane Labs

For CTOs and eng leads rolling out AI-assisted dev without quality regressions.
For teams shipping agents that need evals, guardrails, and observability.
For protocol teams in audit or post-launch hardening.

Book a Discovery Call

Services

Choose a focused sprint.

Clear deliverables. Measured outcomes.

Agentic Integration Sprint

2 weeks

Ship an agent into real workflows safely.

  • Integration with repos/services (PRs, CI, tickets, runbooks)
  • Eval harness + golden tasks + regression checks
  • Guardrails (permissions, tool boundaries, fail-closed)
  • Observability (traces, cost, latency, failure modes)

Success: Measurable quality + reliability, not demos.

AI Upskilling & Enablement

2–4 weeks

Make the team faster with AI without regressions.

  • Baseline workflow audit (where AI helps / harms)
  • Team training + live pairing (Claude Code / Codex)
  • Repo templates: prompts, policies, review checklists
  • Measured before/after (cycle time, CI cost, defect rate)

Success: Adoption sticks + quality maintained.

Protocol Hardening

1–3 weeks

Audit remediation, test upgrades, and release hardening.

  • Post-audit remediation PRs with full test coverage
  • Invariant and fuzz testing upgrades
  • Security hardening for Solidity + backend services
  • Release checklist and launch engineering

Success: Audit findings resolved + deployment-ready.

Track Record

Recent work.

FastLane Labs

View

Problem: CI pipeline costs scaling unsustainably alongside protocol complexity.

What I did: Restructured CI/CD, shipped LST protocol core, handled post-audit remediation, and built multi-region platform infrastructure.

Cut CI spend ~70%. Protocol shipped to mainnet.

FastLane Labs

Problem: Engineering team manually reviewing PRs and writing boilerplate without AI assistance.

What I did: Led Claude Code / Codex CLI rollout across the team. Built context engineering templates, review checklists, and agent-assisted workflows.

Team-wide AI adoption with measurable quality improvements.

Squidrouter

View

Problem: No unified cross-chain swap infrastructure across EVM and Cosmos ecosystems.

What I did: Co-founded and engineered the cross-chain swap protocol spanning 13+ EVM and 50+ Cosmos chains.

$500M+ transaction volume. Exited.

Selected tools: Claude Code · Codex CLI · Context Engineering · Prompt Design · Multi-agent Orchestration · Tool Use · RAG Pipelines · MCP · Anthropic API · OpenAI API · Agent SDK · Eval Frameworks · Solidity · Foundry · EVM · Cosmos SDK · Go · TypeScript · Python · Node.js · AWS (EKS) · Kubernetes · Terraform · CI/CD

Process

How engagements work.

01

Discovery

30-minute call

Understand your stack, goals, and constraints. Identify the highest-impact starting point.

02

Fixed-Scope Sprint

1–4 weeks

Hands-on delivery against a defined scope. You get weekly demos, written updates, and repo PRs.

03

Rollout & Handoff

End of sprint

Documentation, team training, and a clean handoff. You own everything.

Contact

Let's talk.

  • What are you trying to ship in the next 2–4 weeks?
  • Where should an agent plug in (repo/CI/support/oncall)?
  • Any constraints (security, data, stack)?

Prefer a conversation?

Book a Discovery Call