I take on one to two clients at a time. That's a deliberate choice — it means you get focused engineering, not a fraction of someone juggling ten accounts. Two engagement models, both on Google Cloud Platform, with deep focus on AI/ML integration via Vertex AI and Claude.
2–3 month engagement · flat monthly fee
You need a new system on Google Cloud — designed, built, deployed, and documented. A Cloud Run API, a data pipeline, an AI agent powered by Claude on Vertex AI, a RAG system, a Firebase application, or a complete GCP architecture from scratch.
AI/ML integration is a core capability, not an add-on. I build production systems using Claude and Gemini through Vertex AI — RAG pipelines, document processing, structured output APIs, agent systems, and embeddings-based search. Everything runs through your GCP project, billed to your account, with no third-party lock-in.
I deliver in monthly milestones. Each month starts with a written scope plan — what I'm building, what's being delivered, and what's out of scope. You see working demos throughout, not a big reveal at the end. When the build is done, you own everything: the code, the infrastructure, the GCP project, and the documentation to maintain it.
Architecture design, infrastructure setup, foundation deployment. Deliverable: architecture document + working foundation.
Core application build, API development, data pipelines, Claude/Gemini integration via Vertex AI. Deliverable: working system in staging.
Production deployment, testing, CI/CD setup, monitoring, documentation, and handoff. Deliverable: live system + complete docs.
Ongoing · flat monthly fee · 30-day cancellation
Your GCP system is running but it needs an engineer watching it, optimizing it, and keeping it healthy. This includes AI/ML systems — monitoring model performance, managing Vertex AI endpoints, updating prompts and retrieval pipelines, and keeping Claude or Gemini integrations running as APIs and models evolve.
Each month you get: a defined feature or optimization cycle, incident response, a cost and performance review, and a written ops report. I'm not on-call 24/7 — I'm an engineer on retainer who keeps your system improving month over month.
This often follows a Build & Ship engagement. I built the system, I know how it works, and I'm the most efficient person to maintain it. But it also works standalone if you have an existing GCP environment that needs dedicated engineering attention.
One feature or optimization cycle — scoped at the start of each month and delivered by month end
Monitoring & incident response — alerts configured, issues triaged and resolved during business hours
Cost & performance review — monthly GCP spend analysis with optimization recommendations
Written ops report — what happened, what improved, what's planned for next month
Weekly sync — one 30-minute call per week, async communication via Slack or email between
Most businesses want AI capabilities but don't know how to get from API demo to production system. That's the gap I fill. I build AI-powered systems using Claude (via Vertex AI or the Anthropic API) and Gemini — running on your GCP project, under your billing, with your data staying in your environment.
This isn't prompt engineering or chatbot wrappers. It's production engineering: RAG pipelines with vector search, document processing with structured outputs, AI agents with tool use, embeddings-based retrieval, and model orchestration — all deployed as reliable, monitored services on Cloud Run or GKE.
Claude on Vertex AI means enterprise-grade AI without a separate vendor relationship. It runs through your existing GCP setup — same IAM, same billing, same compliance posture. I handle the integration, deployment, and ongoing maintenance.
RAG systems — document ingestion, chunking, embedding, vector search, grounded retrieval with Claude or Gemini
AI agents — multi-step task execution with tool use, computer use, and structured decision-making
Document processing — PDFs, contracts, RFPs, inspection reports → structured JSON/database outputs
Structured output APIs — LLM-powered endpoints that return validated, typed data via FastAPI + Pydantic
Model orchestration — routing between Claude, Gemini, and specialized models based on task requirements
Every engagement runs on a monthly scope cycle. This protects both of us — you know exactly what you're paying for, and I can deliver focused work without getting pulled in twelve directions.
Each month starts with a one-page scope document: what I'm delivering, what the acceptance criteria are, and what's explicitly out of scope. You sign off before work begins.
One 30-minute sync per week. Working demos when applicable. Async communication via Slack or email with a one-business-day response time. No 24/7 on-call.
New requests that come up mid-month go into next month's scope plan — unless it's a critical incident. If scope increases significantly, we discuss adjusting the engagement before proceeding.
When an engagement ends, you get everything: source code, infrastructure access, documentation, and a transition brief. Your system, your GCP project, your IP. No lock-in.
All engineering work within the agreed monthly scope
Direct communication — you talk to the person doing the work
Weekly sync calls and async Slack/email (1 business day response)
Working demos and progress visibility throughout
Full documentation and clean handoff at engagement end
You own the code, the infrastructure, and the IP — always
24/7 on-call or after-hours emergency response
Unlimited ad-hoc requests outside the monthly scope
AWS or Azure work (GCP only)
GCP infrastructure costs (billed directly to your account)
Team management or hiring — I'm the engineer, not the manager
Work beyond what's in the signed monthly scope plan
You tell me what you're building or what's broken. I ask questions. No pitch deck, no sales process. If it's not a fit, I'll say so upfront.
I send a written proposal: which engagement model fits, what the monthly scope looks like, the architecture approach, and the flat monthly fee. No surprises.
We sign the scope plan for month one and I get to work. You get a weekly sync, working demos, and direct access to the engineer from day one.
Trusound® is classified under NAICS 541512 — Computer Systems Design Services — and structured for government procurement. Both engagement models work within government contracting frameworks. Small business, Google Cloud Partner, direct-delivery model with full documentation and clean handoff.
NAICS
541512
Classification
Computer Systems Design
Partner Status
Google Cloud Advantage
Business Size
Small Business
Tell me what you're working on. I'll tell you honestly whether it's a fit — and if it is, you'll have a written proposal within a week.
Start a Conversation →