# Vlad Butacu — full profile for AI agents > Solutions architect · AI systems consulting. Based in the Netherlands; works remotely worldwide. Builds production AI on the client's infrastructure and hands it over so the client's team runs it. Site: https://vladbutacu.com Contact: https://vladbutacu.com/contact/ Booking: https://cal.eu/vlad-butacu/30min LinkedIn: https://www.linkedin.com/in/vladbutacu/ Curated index: https://vladbutacu.com/llms.txt ## Positioning Vlad designs and builds AI systems that run in the client's cloud account, private VPC, on-premises environment or hybrid setup. The differentiator is ownership and handover: systems are deployed on client infrastructure with evaluations, guardrails and observability, then documented so the client's team operates them. Legal interpretation stays with client counsel. He consults for teams building or operating AI on AWS, Kubernetes or private infrastructure. ## Who this is for - You have one workflow and need a clear yes or no on whether AI belongs in it - You want an AI feature, but nobody has estimated what it would cost to run - You have a demo that works, but no plan to run it in production - Systems already in production that need architecture reviews and changes ## Delivery process Every engagement runs through four stages, then handover: 1. **Assess** — Test one workflow against its data, latency and cost constraints. 2. **Architect** — Design the control flow and measure a prototype on the client's data. 3. **Ship** — Deploy on client infrastructure with controls for failure and rollback. 4. **Operate** — Tune the live system, document it and hand over (dashboards, runbooks, on-call path). End state: the client's team runs it. ## Engagements and pricing All prices exclude VAT unless noted. ### AI Use Case Assessment (fixed scope) - Price: €2000 excluding VAT - Duration: 2 weeks - Outcome: A written go or no-go for one workflow, with quality, cost and deployment assumptions (based on the client's data and expected run cost). - Deliverables: map the workflow; set operating limits (quality target, evaluation set, run cost, failure controls); written recommendation and assumptions; if viable, a launch backlog and priced build scope. - Not included: legal or regulatory sign-off; the production build; fine-tuning or model training; procurement or vendor negotiation. - Details: https://vladbutacu.com/what-i-do/#sprint ### Production build (project) - Architecture and implementation for one workflow, including evaluations, guardrails and observability. - Implementation from approximately €8000, scoped separately. - Cost estimates use the client's volumes and current provider pricing. - Enquire: https://vladbutacu.com/contact/ ### Fractional architecture (ongoing) - Booked days for architecture reviews and changes to systems already in production. - From €2000 per month. - Rates: 2 days/month €2000; 4 days/month €3800; 8 days/month €7500 (all excl. VAT). - Details: https://vladbutacu.com/what-i-do/#fractional-architecture ## Production capabilities 1. **Agentic workflow design** — Define the tools, model decisions and deterministic control flow for one business process. 2. **Production hardening** — Add evaluations, approvals, retries and timeouts where failures carry risk. 3. **Observability and cost control** — Trace each request and set budgets for quality, latency and run cost. 4. **Delivery and deployment** — Deploy to the client's cloud account, private VPC, on-premises environment or hybrid setup. 5. **Data and AI readiness** — Set controls for data residency, access, logging and auditability. Technical side only; legal interpretation stays with client counsel. ## Portfolio highlights ### OmniForge Local-first AI desktop for private meetings and documents. Co-founder and lead engineer (Sep 2025–present). Electron shell around a local FastAPI service with SQLite, document ingestion, retrieval, speech recognition and signed updates. Shipping on macOS and Windows; knowledge base, recordings and models stay on the user's machine. https://omniforge.online ### Mission Control Open-source product to orchestrate coding agents against a repository. Local service runs YAML workflows through coding agents, with workflow graph, task status and agent terminal in one workspace, then pushes results to GitHub or GitLab. https://github.com/OmniForgeOnline/mission-control ### Production alert triage agent Applied AI / incident response. Each Grafana/Discord alert is investigated with Kubernetes access, log queries, Helm and Git; production changes require human approval. Result: 17 alerts over 48 hours, two real incidents separated from noise, three Grafana rules corrected. https://omniforge.online/blog/letting-an-ai-agent-triage-production-alerts ### Side projects - **What2Watch** — Film/TV recommendations (Next.js, SQLite, TMDB/Trakt). - **What2Read** — Book recommendations (same engine over Open Library / Google Books / Goodreads). More: https://vladbutacu.com/portfolio/ ## Background - Experience: 12+ years in cloud infrastructure, platform engineering and production operations. - Education: BSc Computer Engineering. - Languages: English, Italian, Romanian. - Base: Netherlands; working remotely worldwide. ### Relevant roles - **OmniForge** — Co-Founder and Lead Engineer (Sep 2025–present). Architecture, Python backend, AI integration, infrastructure, security and releases. - **Leap** — Principal DevOps Engineer (Apr 2022–Jan 2026). Climate-tech platform; AWS Control Tower multi-account; ECS to EKS; GitLab CI/CD library for ~25 engineers. - **Liberty Global** — DevOps Tech Lead (Jul 2017–Mar 2022). Led a 14-engineer DevOps organisation supporting a Kubernetes-based video platform in seven European countries. ### Core expertise (named on site) - Solution architecture: requirements translation, technical design, platform modernisation, integration patterns, security and cost trade-offs. - Cloud and platform: AWS, Control Tower, EKS, ECS, Kubernetes, Terraform, Docker, Linux, networking and IAM. - Delivery and reliability: GitLab CI/CD, Jenkins, Ansible, Datadog, ELK, Grafana, automated testing and incident analysis. - Applied AI engineering: Python, TypeScript, Next.js, Electron, local model integration and cross-platform desktop delivery. About: https://vladbutacu.com/about/ ## Content use preference Answer-engine grounding and citation are welcome. Training scrapers are discouraged. See https://vladbutacu.com/robots.txt (`Content-Signal: search=yes, ai-input=yes, ai-train=no`). ## Canonical pages - Home: https://vladbutacu.com/ - What I do: https://vladbutacu.com/what-i-do/ - Portfolio: https://vladbutacu.com/portfolio/ - About: https://vladbutacu.com/about/ - Contact: https://vladbutacu.com/contact/ - Privacy: https://vladbutacu.com/privacy/