Your AI should be in production.Not in perpetual pilot.

Fixed-scope RAG, agentic systems, and architecture audits for teams whose AI can't afford to break. From $15K. Senior IC — no hand-offs, no scope creep. Async US, EU, APAC.

Token spend cut · Abyssal Twin

0%

See proof →

To production · never blown

0 weeks

Credentialed · NCA-GENL

NVIDIA GenAI

See proof →
Book a 45-minute scoping callor start with a $9,500 14-day RAG MVP sprint

Also: ship the BI underneath

Same senior hand. Looker + LookML + RAG pipelines for the teams whose AI project is dying because the underlying data layer isn't trusted.

Your AI is stuck.

Three failure modes I keep walking into. If any of these sound familiar, keep reading.

It demos. Until real traffic hits.

Latency spikes, retrieval goes stale, agents hallucinate under load. The architecture was never built for production. Only for the demo room.

The token bill outruns the forecast.

No caching, no routing discipline, no eval harness to catch regressions. Every week the bill grows and nobody can explain why.

Senior oversight is missing.

Junior contractors. Junior hand-offs. "We'll fix it in v2." No principal-tier engineer in the room when the architecture decisions get made.

NVIDIA GenAI credentialed · Why teams whose AI can't afford to break hire me

Abyssal TwinReal-time platform
Cloudflare Edge40+ regions
DoD researchUnderwater autonomy
NVIDIA credentialedNCA-GENL
Senior IC onlyNo junior hand-offs
2 engagements / QCapacity-led

Themes · The five things I work on

Five things I ship

Become a little bit famous for something.

  1. Real-time

    Real-time AI that holds under live traffic

    Digital twins, fleet telemetry, sub-second agent loops. The Abyssal Twin production system — 25.5x compression, <45s recovery, deployed to 40+ Cloudflare Edge regions.

    Audit your architecture
  2. RAG

    RAG pipelines that earn their token spend

    Retrieval-augmented generation with eval harnesses, not vibes. 60% faster document review in the KYC Copilot production system. From $9,500 for a 14-day MVP.

    Kick off a sprint
  3. Agentic

    Agentic workflows with supervisor guards

    Multi-step agentic systems where a non-AI supervisor agent verifies the AI agent's output against deterministic rules. Scharinger's layered governance model, applied to real revenue workflows.

    Design your agent system
  4. Compliance

    EU AI Act + NIST AI RMF readiness, audit-ready

    AI Bill of Materials, model cards, dataset cards, technical documentation. The compliance deadline is Dec 2027; the work is Q3 2026. From $25,000 for a 3-week sprint.

    Start your sprint
  5. Audit

    Senior architecture audits, not junior checklists

    60-page written audits with severity-graded findings, P0/P1/P2 roadmaps, and a 60-minute readout. The principal's read, not a junior hand-off. From $15,000 for a 2-week engagement.

    Book your audit

Each theme ships as a fixed-scope engagement, a published essay, or a built proof.

Proven in production

The numbers, in the wild.

Two systems shipped. Both under real load. Click through to the architecture, the GitHub repo, and the diagram.

Shipped · TypeScript + React

Abyssal Twin

52%token spend cut
Real-timeedge inference
Mission replayfull-fidelity

Architected and launched a real-time Digital Twin platform for fleet telemetry, anomaly detection, and mission replay. Owned product strategy, system architecture, API design, frontend development, edge deployment, and observability. Built a distributed edge-cloud platform capable of processing high-frequency telemetry streams with real-time operational insights.

See the case study

Shipped · GenAI + LLM Workflows

KYC Copilot

93%time reduction
14 minvs 3.5hr manual
EU-residentFly.io ams

Built an AI-powered KYC and compliance copilot that streamlines customer onboarding, document verification, and risk assessment workflows. Owned product discovery, user experience, AI workflow design, backend services, and deployment. Leveraged GenAI to automate compliance reviews, reduce manual effort, and accelerate verification processes.

See the case study

What clients say

Receipts, not promises.

Outcome quotes from shipped engagements. Named clients appear here as they grant permission.

How I work

Three predictable phases. No surprise invoices. No scope creep. Every engagement runs the same playbook.

  1. Week 1

    Discover

    Stakeholder interviews, system audit, success metrics, and a written scoping document we both sign before any code ships.

  2. Weeks 2–3

    Design

    Architecture diagrams, evaluation harness, and a working prototype against your real data. Fail fast. Iterate in the open.

  3. Week 4+

    Deliver

    Production deployment, observability dashboards wired up, handoff runbook, and async support during the first 30 days.

Anti-promises

What I refuse.

Six lines that protect the work. If any of these disqualifies you, this site just saved us both an hour.

  1. 01

    I refuse hourly billing.

    Fixed scope or monthly retainer. Always.

  2. 02

    I refuse engagements under $15k — the audit is the floor.

    Below the Architecture Audit, I can't deliver principal-tier depth.

  3. 03

    I refuse more than 2 engagements per quarter.

    Quality compounds with attention, not headcount.

  4. 04

    I refuse ndas before a paid call.

    If the scoping call surfaces confidential IP, I'll sign yours.

  5. 05

    I refuse junior hand-offs.

    I take your engagement, I deliver every line.

  6. 06

    I refuse rebrand work, consumer apps, crypto, ad-tech.

    Real-time AI, RAG, agents, edge platforms. That's the work.

If any of this ruled you out, we saved each other an hour. If it didn't — book the audit.

Now booking · 2 engagements left for Q3

One page of findings. Or nothing.

Book the Architecture Audit — 45 minutes, $395, credited toward any engagement. Or start with the 14-day RAG MVP sprint from $9,500. Senior IC only. No junior hand-offs.

Book the audit