Technology leadership & architecture — Malmö, Sweden

Micael Carlstedt Consulting

Senior technology leadership, architecture and AI-assisted engineering.

Micael Carlstedt Consulting shapes technology vision, strategy and architecture, then stays close to the code and the team to ship it. Pragmatic, people-first and quality-driven.

Strategy Code Works at both altitudes.

Start a conversation

01 About

Two altitudes, one partner.

Micael Carlstedt Consulting is a senior technology-leadership and architecture practice, part of Intersum. It works at two altitudes — shaping technology vision, strategy and architecture, then staying hands-on with the code and the team to ship it. Structured and pragmatic: technology aligned to business goals, empowered teams, and people ahead of process while the work still gets done.

Defense
Building automation
Finance
Telecom
Retail

At the strategic level

Shapes the technology vision and strategy in step with business goals
Owns the high-level architecture across teams and products
Balances technical risk against operational cost
Grows healthy engineering organizations and cultures

Close to the code

Works hands-on across system and solution architecture
Joins the ceremonies the team owns — dailies, reviews, retros, demos
Mentors developers and sets the quality bar
Stays a technical lead, not a detached architect

02 Services

Three ways MCC helps a team.

Technology leadership & architecture

Set the technical direction, then stay close enough to the code to make it real. MCC works at both altitudes — shaping vision, strategy and architecture across teams and products, and leading hands-on as a technical lead.

Technology vision and strategy aligned to business goals
Cloud-native, event-driven architecture with Domain-Driven Design
Microservices, Event Mesh and infrastructure as code
Balancing technical risk against operational cost

DDD
Microservices
Event-driven
TypeScript
Go
.NET / C#
GCP
Terraform

AI-assisted & agentic engineering

Bring modern AI into a team as a governed collaborator, not autocomplete. MCC designs an AI assistance network that amplifies delivery while keeping engineering quality and human judgement firmly in control.

Reusable agent skills that encode a team's golden-path workflows
Multi-agent orchestration: explore, plan, implement, adversarially verify
Custom instructions, persistent memory and MCP wiring
Pragmatic governance — generated code is a proposal to be verified, never shipped blind

Claude Code
GitHub Copilot
Agent skills
Subagents
MCP
Prompt/context engineering

Quality, delivery & ways of working

Build quality in from the start and keep a steady flow of delivery. MCC introduces the engineering practices and the agile ways of working that let empowered teams own their craft and ship continuously.

TDD/BDD and automated quality gates for lasting maintainability
Full CI/CD, ephemeral preview environments, progressive delivery
DevSecOps shifted left — fail-fast scans on every pull request
Coaching teams toward autonomy; measuring with DORA

TDD / BDD
CI/CD
GitHub Actions
DORA
DevSecOps
Trunk-based

03 AI-assisted engineering

An AI assistance network, kept under human control.

MCC treats modern AI as a team of collaborators rather than autocomplete. Working fluently across the leading agentic coding tools — primarily Anthropic Claude (Claude Code) and GitHub Copilot — the practice designs an AI assistance network that amplifies delivery while keeping engineering quality and human judgement firmly in control.

Agent skills

Authoring and packaging reusable skills that encode domain knowledge and golden-path workflows, so the same expertise is applied consistently across a team.

Agents & subagents

Fanning work out to parallel, specialised agents — explore, plan, implement, adversarially verify — and synthesising the results into multi-agent workflows for large migrations, audits and reviews.

Custom instructions

Steering agents with precise, version-controlled project instructions so output matches the team's conventions rather than generic defaults.

Memory

Persistent, repository-scoped knowledge that lets agents retain patterns, decisions and preferences across sessions.

Model Context Protocol

A tool-agnostic standard that connects agents to repositories, issue trackers and internal systems.

Pragmatic governance

Clear guardrails for when AI is trusted and when a human reviews. Generated code is treated as a proposal to be verified, never a result to ship blindly.

MCC treats AI the way it treats people. A new developer is onboarded with clear instructions, conventions and ways of working, then grows from guided to autonomous along a maturity curve — and the same applies to an agent. Instructions, skills and agents “onboard” it so it grasps the team's context, domain and process, and consistently delivers high-quality, review-ready output.

04 Leadership & ways of working

Empowered teams, steady flow.

MCC's way of working is rooted in the Agile Manifesto: individuals and interactions over processes and tools, and building around motivated people given the environment, support and trust to get the job done. Rather than running the day-to-day, it guides the team to build this way of working together — empowered, autonomous and self-organizing — removing impediments and creating the conditions for ownership, steady flow and continuous improvement.

A transparent board the team owns

Backlog
To do
In progress
In review
Ready for review
Acceptance testing
Done

Empowered, self-organizing teams

Built around motivated people who are trusted, with the support to get the job done. Decisions are made by the people closest to the work, not imposed from above.

Guidance over direction

Coaches and clears the path — asks the right questions, removes blockers — instead of assigning and supervising. The team owns its way of working.

Pull top-left, limit WIP

Start the highest-priority work first and finish before starting more, rather than everyone starting their own thing.

Deliver early and often

Frequent delivery of working software, so quality is built in from the start and carried through continuous delivery, not inspected in at the end.

Own your story to Done

Whoever picks up a ticket owns it the whole way, through review and acceptance. The team swarms on blockers and helps each other finish.

No framework dogma

Kanban, Scrum or a blend — the team adopts the practices and cadence that fit the work. People over process, and the same principles hold as teams grow.

CI/CD: continuous delivery, deploy on green
DORA: all four delivery metrics
PR: ephemeral preview per pull request

06 Capabilities

The engineering data sheet.

Languages: TypeScript / JavaScript
Go
C#
Python
C++
Java
Cloud & platforms: Google Cloud (GCP)
Cloud Run
Alibaba Cloud
Docker
Microsoft Azure
AWS
Architecture: Domain-Driven Design
Microservices
Event-driven / Event Mesh
Serverless
Micro-frontends (ESI)
Enterprise Integration
APIs & integration: REST
OpenAPI
AsyncAPI
KrakenD gateway
Event brokers
Bruno
Data & storage: Cloud SQL / PostgreSQL
BigQuery
Datastore
MongoDB
Neo4j
SQL Server
Frontend: React
Vite
Micro-frontends (ESI)
Astro
Tailwind
Delivery & IaC: Terraform
GitHub Actions
CI/CD
Feature flags
Ephemeral previews
DORA metrics
Trunk-based
DevSecOps: CodeQL (SAST)
Dependabot
Renovate
Trivy
JIT least-privilege
Automated key rotation
Observability: OpenTelemetry
Distributed tracing
Grafana
Cloud Monitoring & Logging
AI-assisted engineering: Claude Code
GitHub Copilot
Agent skills
Subagents & orchestration
MCP
Prompt / context engineering
Testing: Playwright
Vitest
Jest
Contract testing
xUnit / nUnit
SpecFlow (BDD)
Methods: Agile
Scrum
Kanban
TDD
BDD
DDD

07 Contact

Have a complex systems effort that needs steering and shipping?

Available for senior technology-leadership and architecture engagements through Micael Carlstedt Consulting AB. The quickest way to start is an email.

micael@carlstedt.com

Full CV and a detailed project history are available on request, via a private access code.

Elsewhere: LinkedIn GitHub
Based in: Malmö, Sweden