Kai HeldSITE RELIABILITY ENGINEER
Contact

Profile

Kai
Held

I fix things you didn’t even know were broken.

I’m a Technical Delivery Lead and Site Reliability Engineer in a permanent role at Helsing. Independently, I take on selected part-time or short-term work around secure infrastructure, private-cloud platforms, and operational reliability.

Send context

Principles

How I operate

01

Make it work

Start with the smallest solution that proves the path. I value delivery over speculative abstractions: no overengineering, premature optimization, or perfectionism that prevents useful systems from shipping.

02

Make it right

Turn the working path into a dependable system. Build in reliability, security, compliance, observability, maintainability, and clear ownership — especially in high-assurance environments.

03

Make it fast

Measure where time and capacity are actually lost, then optimize. Improve latency, throughput, resource efficiency, delivery lead time, and recovery speed — without trading away correctness.

Point of view · OSI 01—07

The full-est stack.

Own the whole system
— not only the code that happens to be visible.

“Full-stack” often stops at front end and back end. That is a useful distinction, but an incomplete system model. In the age of AI, leverage at the upper layers is growing fast; the dependencies underneath are not getting simpler. Reliability still emerges from the entire chain, down to networks, hardware, power, and the physical environment.

System view

USER ↕ REALITY
  1. 07ApplicationInterfaces & behavior
  2. 06PresentationEncoding & exchange
  3. 05SessionState & coordination
  4. 04TransportDelivery & flow
  5. 03NetworkRoutes & reachability
  6. 02Data linkFrames & adjacency
  7. 01PhysicalPower, signal & hardware

01—04 Where a neat architecture diagram becomes a dependency graph.

Experience

Work record

Current position

Permanent role · since February 2025

  1. 01

    Helsing

    Technical Delivery Lead / Site Reliability EngineerPermanent position

    Building and delivering highly available, secure infrastructure in defense technology.

Independent consulting

Full-time self-employed · December 2022 — January 2025

  1. 01

    BMW Group

    AWS DevOps EngineerIndependent contract

    Enabled internal teams through robust cloud platform capabilities on AWS.

  2. 02

    Bundesdruckerei-Gruppe

    Site Reliability EngineerIndependent contract

    Built and operated a data science platform for government use in a highly secure environment.

  3. 03

    Rohde & Schwarz

    Backend Developer / PythonIndependent contract

    Developed identity and access management capabilities for a high-security environment.

  4. 04

    InformMe

    Senior AWS DevOps EngineerIndependent contract

    Architected a GDPR-conscious AWS account factory for a MedTech company.

  5. 05

    Volkswagen

    Senior AWS Security ArchitectIndependent contract

    Supported AWS IAM security and cloud identity management.

Earlier employment

ventx GmbH · employee · 2020 — 2022

  1. 01

    Eurofins

    DevOps EngineerClient assignment via ventx

    Architected and engineered a greenfield AWS EKS platform and its supporting services.

  2. 02

    Siemens

    DevOps EngineerClient assignment via ventx

    Migrated an on-prem learning platform to AWS and supported the Siemens Energy carve-out.

Systems

Capability stack

PLATFORMS
AWS · EKS · Kubernetes · Linux
SOVEREIGNTY
On-prem Kubernetes · Bare-metal bootstrap · Self-hosted storage · GitOps
IDENTITY & PKI
Keycloak · OIDC/SAML · PKI · Secrets & key management
ASSURANCE
BSI IT-Grundschutz · NIST · ISO/IEC 27001 · NATO-aligned constraints

Education

Foundations

2017—20

Friedrich-Alexander-Universität

Master of Arts · Philosophy

MAR 2016

Max Weber Programme Elite Scholarship

Free State of Bavaria · Scholarship

2012—17

Otto-Friedrich-Universität

Bachelor of Arts · Philosophy & Political Science

Working set

Things I like to work with

01Shared thinking

Whiteboard

The original multimodal interface: fast, shared, and gloriously low-latency. If we cannot draw the system together, we probably do not understand it together.

02Kept on a short leash

AI

Agents earn their keep by shortening the distance to a useful answer. I still read every diff and deny the ones that go sideways.

03Still in the loop

AI (Actual Intelligence)

A model gives you a conclusion. It cannot give you a reason to believe it, only a pattern that resembles one. That part is still mine.

Selected work

A concise note
is enough.

[email protected]
Environment · constraint · timing

Send context