Skip to content
Dominik Heinrich

Dominik Heinrich, multi-cloud DevOps & AI solution architect

I make cloud platforms cheaper, calmer and faster to ship on.

I build and run secure cloud platforms on Azure, AWS and Kubernetes and cut what they cost. And I make companies AI-native with secure AI platforms and agents that take over the manual work, so small, mid-size and large teams run lean, fast and ready for what's next. 7+ years across banking, automotive, public sector and SaaS.

Available worldwide, remote, and on-site on request

Cloud cost per month

Weeks

Tap a step on the chart to see what I do there.

Real result from a recent engagement: total Azure costs down 48% in five weeks. See the case

Illustration: how a cloud bill typically comes down in a FinOps engagement

How FinOps works

Cost work is a routine, not a one-off clean-up. Five steps, and what you get to see in each one.

See the cost

Nobody saves what they can't see. First everyone gets the same picture: cost per month, per team and per service, with an owner for every line.

  • Amortized cost per resource from the billing API, not the invoice PDF
  • Tags for product, team and stage; untagged spend gets an owner within a week
  • A run-rate forecast, so a change shows up in days, not at month end
€ / monthTap a month
FinOps starts
MarAprMayJunJulAugSepOct
Compute
1.4k
Databases
0.8k
Storage & logs
0.4k
Network & other
0.3k
Total
2.9k €

Illustration with rounded numbers, shaped like a real engagement.

Services

What I build and run for you, from the cloud bill to the finished app. For startups, mid-sized companies and enterprises, always tailored, on my own or with a trusted senior team for larger projects.

  • AI-native operations

    I map the manual work in your company and hand it to AI agents and automations: inboxes, reporting, sales follow-ups, testing. People keep the approvals, the team gets its time back.

    Learn moreSee example

  • AI agents for WhatsApp, phone and chat

    Assistants that answer your customers on WhatsApp, on the phone and in chat, learn from your team's answers and hand over to a person when it matters.

    Learn moreSee example

  • StillAnswer: AI assistant for WhatsApp, set up for you

    Answers your customers on WhatsApp 24/7 from your knowledge base, looks up bookings and orders in your tools, alerts your team when it can't answer and learns from their replies. Staff inbox included. Your accounts or all-inclusive.

    Learn moreSee example

  • Production AI platforms (LLMOps)

    Running AI like any other critical system: evaluation suites and regression tests, controlled tool access through MCP with least privilege, identity, tracing, rollback, and cost and latency budgets.

    See example

  • Local & private AI for sensitive data

    Open models on your own hardware or private cloud: the right model and quantization for the job, retrieval on your documents, personal data anonymized first. Nothing sensitive goes to a public API.

    Learn more

  • AI integration

    AI agents, chatbots and MCP servers connected to your systems, with a confirm step and human approval where it matters.

    Learn moreSee example

  • Automation of recurring tasks

    Reports, data syncs, checks and hand-offs your team repeats every week, run by scheduled jobs and AI agents, with an alert when something fails.

    See example

  • Automated prospecting & lead generation

    Prospects found and scored automatically, personal outreach and follow-ups drafted by AI, and a person approves before anything is sent. With a pipeline and a sales report for management.

    See example

  • GEO & SEO: found in Google and in AI answers

    How ChatGPT, Gemini, Perplexity and Google's AI Overviews see your brand today, measured with Semrush and Peec AI. Prompt research at scale with open models such as Qwen, run privately. Then structured data, citable content and listings, fixed in order of impact.

    Learn moreSee example

  • Cloud cost optimization

    Cost baseline per resource, right-sizing on real peak load, reservations and budgets with forecast alerts. Savings without anything your customers would notice.

    Learn moreSee example

  • License & SaaS spend overview

    An inventory of every license, tool and subscription: who uses what, what overlaps, what renews when and what you can cancel.

    Learn more

  • Kubernetes orchestration & optimization

    AKS, EKS and Talos on bare metal, VMs on KubeVirt, GitOps with FluxCD and Helm. Right-sized on real p95 load, fewer nodes, autoscaling with Karpenter.

    Learn moreSee example

  • Cloud platforms & IaC

    Azure and AWS foundations in Terraform or Bicep, with strict Dev, Test and Prod separation. Secure, reproducible and easy to roll back.

    Learn moreSee example

  • Cloud, on-prem or hybrid

    The right place for each workload: public cloud, your own data center or both, connected and run as one platform, air-gapped where needed.

    Learn moreSee example

  • Legacy migration

    From old servers, monoliths and on-prem databases to the cloud or Kubernetes: AI-assisted code analysis, a plan in steps, parallel runs and cut-overs with identical row counts and checksums.

    Learn moreSee example

  • Enterprise security & scaling

    OIDC and managed identities, Entra ID, Azure Policy and RBAC, Key Vault and External Secrets, row-level security and encrypted backups, on platforms that scale with autoscaling and tenant isolation. Used in banking, automotive, the public sector and SaaS.

    Learn moreSee example

  • Product development

    From idea to MVP to a product customers pay for: scope, architecture, payments, onboarding and a release rhythm that keeps improving it.

    See example

  • App development

    Full-stack apps with React, TypeScript, Supabase and Stripe, native iOS and Android when needed. From prototype to production, with staging and tests.

    See example

  • Websites, blogs & web apps

    Fast, multilingual websites and landing pages, deployed automatically on low-cost static hosting.

    See example

  • Embedded & IoT products

    Real hardware products with Arduino, ESP32 and Raspberry Pi: sensors, servos and displays, firmware with PlatformIO, and voice and AI that run fully offline on local models.

    See example

  • CI/CD & release automation

    Pipelines in GitHub Actions and Jenkins, signed mobile releases with Fastlane, preview environments, migrations checked before merge and smoke tests after every deploy.

    See example

  • Monitoring, alerting & reliability

    Prometheus, Grafana and Loki plus outside-in checks that alert on business outcomes, not just CPU. Backups restore-tested every night.

    See example

Ready-made products

Two products I built once and set up for clients again and again, so you get a proven system instead of a project from zero.

Live on this site

Helptempo

AI chat for your website

A sales and support chat for any website. It answers from your own knowledge, sends visitors to the right page and suggests the next step. The guide at the bottom right of this page runs on the same engine.

  • One line of code on your site
  • Quotes only the facts and prices your site shows
  • Answers stream in while they are written
  • English, German, Spanish and French

Can you help us cut our cloud bill?

Yes. Start with a cost baseline per resource, then right-size on real peak load. In a recent engagement total Azure costs fell 48% in five weeks.

See the FinOps case

Two ways to get it

  1. Hosted: one line of code, access within 2 weeks
  2. In your own stack: I set up the same engine on your hosting, with Gemini, Claude, an OpenAI-compatible API or local models

How to get it: send a short request or book a call. You get a fixed offer within 1–2 business days.

See the caseProduct site: helptempo.com
In production for a hospitality business

StillAnswer

AI assistant for WhatsApp

Answers your customers on WhatsApp around the clock, in their language, from your own knowledge base, and hands over to your team when it matters. For surf schools, hostels, retreats, training providers and shops.

  • Takes action through your tools: looks up a booking or order, checks availability, records a change request
  • Knows its limits: tells the customer you'll check and alerts your team by email and WhatsApp with a link to the chat
  • Learns from your team: unanswered questions become suggestions, your reply is pre-filled as the answer, one click to teach it
  • Staff inbox: read every chat, reply yourself, pause the AI per chat
  • Works with Gemini, ChatGPT or Claude

Two ways to get it

  1. Your accounts: you bring your Twilio and AI keys, I guide the setup
  2. All-inclusive: I provide the number and do the whole setup, including WhatsApp Business approval and message templates

How to get it: send a short request or book a call. You get a fixed offer within 1–2 business days.

More products I build

  • Aulemo: AI voice recorder: recordings turned into transcripts and summaries.
  • Clartelo: Video meetings in the browser.
  • Triggerfold: Embeddable automation engine for your own product.
  • Remoldo: AI website builder: give it a URL, get the site rebuilt.
  • Formwalker: Documents from data: PDFs through a simple API.
  • Bugwalker: AI UX tester that walks through your site and reports bugs.
  • Contentwalker: Content engine for blog posts, social posts and video.
  • Pactwalker: Simple e-signatures for contracts.

Ways to work together

The same quality for a ten-person startup and a global group. The format adapts to you.

Startups

Fractional DevOps & AI lead

A few days a month: architecture, the release pipeline and AI features, so the founders can focus on customers. From prototype to production without a rewrite.

Mid-sized companies

A project with a clear outcome

A defined goal like a lower cloud bill, a migration or the first AI agents in production, with a fixed scope, weekly progress and a handover your team can run.

Enterprises

An architect inside your teams

Embedded with your platform, security and product teams: architecture decisions, governance for AI and regulated environments. For larger scopes I bring a trusted senior team.

How I work

  • Decisions written down: architecture decision records with the trade-offs, so your team knows why, not just what.
  • Secure by default: OIDC and managed identities, least privilege, secrets in a vault, an audit trail.
  • Operated, not just delivered: SLOs, alerts on business outcomes and an incident routine.
  • Measured: cost, lead time, uptime and hours saved, reported every week.
  • Knowledge stays with you: runbooks, docs and workshops for your team.
  • One point of contact who talks to engineering and to management, in German, English and Spanish.

How can I help you?

A few short questions, about a minute. You get a first recommendation right away, and I reply within 1–2 business days.

Step 1 of 4

What do you need?

Choose all that apply.

    Get in touch

    Tell me what you're building. I'll reply within 1–2 business days.

    Dominik Heinrich
    Company
    MADO Consulting

    Available worldwide, remote. On-site on request, depending on the country.

    I reply within 1–2 business days. Free slots: see the calendar

    Opens Calendly's booking calendar here. Loading it sends data to Calendly (see Impressum).