# Contact (/en/docs/contact)



Let's talk about your project.

## Get in touch [#get-in-touch]

* **Email:** [contact@nicoja.hn](mailto:contact@nicoja.hn)
* **Phone:** [+49 175 8989393](tel:+491758989393)
* **LinkedIn:** [nicoja-hn](https://linkedin.com/in/nicoja-hn)
* **GitHub:** [nicoja-hn](https://github.com/nicoja-hn)

I typically reply within one business day.

## For AI tools [#for-ai-tools]

* **MCP server:** `https://nicoja.hn/mcp`
* **llms.txt:** [`/llms.txt`](/llms.txt)
* **Full context:** [`/llms-full.txt`](/llms-full.txt)

**Claude and Claude Desktop** - go to Customize -> Connectors -> + -> Add custom
connector, then enter the MCP server URL above.

**Claude Code** - run:

```bash
claude mcp add --transport http --scope user nicojahn https://nicoja.hn/mcp
```

**ChatGPT** - go to Settings -> Connectors -> Add connector, then enter the MCP
server URL above.


# Privacy policy (/en/docs/datenschutz)



Last updated: 9 September 2026.

## Controller [#controller]

Nico Jahn<br />
c/o COCENTER<br />
Koppoldstr. 1<br />
86551 Aichach<br />
Germany<br /><br />
Email: [contact@nicoja.hn](mailto:contact@nicoja.hn)<br />
Phone: [+49 175 8989393](tel:+491758989393)

## Website delivery and security [#website-delivery-and-security]

This static website is delivered through Amazon CloudFront. Its files are
stored in a private Amazon S3 bucket in AWS region `eu-central-1` (Frankfurt).
Amazon Web Services EMEA SARL and affiliated AWS companies process connection
data, including IP address, timestamp, requested resource, HTTP headers, and
technical browser information, to deliver and protect the website. The legal
basis is Art. 6(1)(f) GDPR; the legitimate interest is reliable and secure
website operation. AWS processes this data on my behalf.

The primary storage and application-processing region is `eu-central-1`.
CloudFront is a global content-delivery network, so technical delivery data may
also be processed at CloudFront edge locations outside the EU/EEA. The AWS Data
Processing Addendum applies to customer data processed through AWS services;
where necessary, its Standard Contractual Clauses and supplementary safeguards
apply. Further information is available in the
[AWS Data Processing Addendum](https://docs.aws.amazon.com/whitepapers/latest/navigating-gdpr-compliance/aws-data-processing-addendum-dpa.html).

CloudFront standard and real-time access logging are not enabled. The website
does not use a separate visitor-profiling database.

## Browser storage [#browser-storage]

The site stores a display preference in local storage only when you select a
color scheme. This is necessary to provide the requested setting (section
25(2) no. 2 TDDDG). It remains stored until you change it or clear your browser
storage. The site uses no optional marketing or tracking cookies and no service
worker for persistent content caching. A technical retirement file removes any
service-worker and cache installation left by an older version of the website.

The AWS delivery components do not use additional browser storage.

<WhitepaperPrivacyDisclosure lang="en" />

## MCP endpoint for AI tools [#mcp-endpoint-for-ai-tools]

The public MCP endpoint makes this website's content available to compatible AI
tools. API Gateway and AWS Lambda process the JSON-RPC request in
`eu-central-1`, in particular its method, request ID, and tool inputs such as
search terms, document paths, language filters, and result filters. AWS Lambda
reads the public documentation from Amazon S3 in the same region. The purpose
is to list, search, and deliver the publicly available documentation.

The legal basis is Art. 6(1)(f) GDPR; the legitimate interest is providing my
public content in a machine-readable form. The MCP endpoint does not store
inputs in an application database, create user profiles, or transmit inputs to
an AI model. API Gateway access logging is not enabled and the Lambda function
is designed not to log request inputs. Technical Lambda runtime logs may be
generated in Amazon CloudWatch to operate and troubleshoot the endpoint. Do not
include personal, confidential, or secret information in MCP requests.

## Contact by email [#contact-by-email]

When you contact me by email, I process your message, contact details, and the
additional information needed to handle the enquiry. This is done to take steps
toward or perform a contract under Art. 6(1)(b) GDPR or, for other enquiries,
under Art. 6(1)(f) GDPR. The legitimate interest is handling business
communication.

The mailbox is provided through Google Workspace, a business email service of
Google Ireland Limited and related Google companies. Processing in the USA may
rely on the EU-US Data Privacy Framework or, where applicable, appropriate
safeguards under Art. 46 GDPR. Further information is available in
[Google's privacy policy](https://policies.google.com/privacy?hl=en).
General email enquiries are deleted 12 months after the last substantive
communication. Statutory retention duties or legal-defense interests may require
longer retention.

## Contact by phone [#contact-by-phone]

When you call, I process data including your phone number, the time and duration
of the call, and the information you provide. My telecommunications provider
processes connection data needed to establish the call. Calls are not recorded.
The legal bases and deletion criteria are the same as for business
communication described above.

## Contact by post [#contact-by-post]

Mail sent to the c/o address stated in the Impressum is accepted, opened,
digitized, and made available in a protected Anschrift.net customer account by
COCENTER GmbH, Koppoldstr. 1, 86551 Aichach, Germany. This processing includes
sender and recipient data, shipment information, and the content of the letter.
COCENTER processes the mail content on my behalf. I receive an email
notification when new mail is available.

The legal basis is Art. 6(1)(b) GDPR for contract-related correspondence and
Art. 6(1)(f) GDPR for other correspondence. The legitimate interest is reliable
receipt and handling of business mail while protecting the private residential
address. According to Anschrift.net, scans remain available until the customer
account is closed or deletion is requested; originals are retained for six
months. Statutory retention duties and legal-defense interests may prevent
earlier deletion. Further information is available in
[Anschrift.net's privacy policy](https://anschrift.net/datenschutzerklaerung/).

Contact by email, phone, or post is voluntary. Without the information required
to handle your enquiry, I may be unable to answer it or take steps toward or
perform a contract.

## Your rights [#your-rights]

Subject to the statutory conditions, you have rights of access (Art. 15 GDPR),
rectification (Art. 16), erasure (Art. 17), restriction (Art. 18), and data
portability (Art. 20). You may withdraw consent at any time with future effect
(Art. 7(3)). Contact the email address above to exercise these rights.

## Right to object [#right-to-object]

Where processing is based on Art. 6(1)(f) GDPR, you may object to that
processing at any time on grounds relating to your particular situation (Art.
21 GDPR).

You may also lodge a complaint with a data protection supervisory authority
(Art. 77 GDPR), in particular in the Member State of your habitual residence,
place of work, or the place of the alleged infringement.

No solely automated decision-making with legal or similarly significant effects
takes place (Art. 22 GDPR).


# How I Work (/en/docs/engagement-model)



I provide consulting and implementation services under a service contract,
billed for actual time at the agreed hourly or daily rate. You work directly
with me — no account managers and no hand-offs.

## 1. Align [#1-align]

We clarify objectives, priorities, constraints, responsibilities, and the
access needed to begin. The agreed activities and initial planning estimate are
recorded in a short statement of work or backlog.

## 2. Book capacity [#2-book-capacity]

We agree an hourly or daily rate, an expected capacity, and a working period.
I invoice the time actually worked, normally monthly and with transparent time
records. Estimates support planning and budget control; they are not fixed-price
commitments.

## 3. Deliver iteratively [#3-deliver-iteratively]

Work proceeds in short feedback loops. You regularly see progress in your
environment and can adjust priorities as we learn. I perform the agreed services
professionally; a particular project result or formal acceptance is not owed
unless separately agreed in writing.

## 4. Transfer knowledge [#4-transfer-knowledge]

Documentation, pairing, and handover are part of the prioritised work. Their
depth follows the agreed capacity and current priorities. Continued support can
be booked when useful.

## Principles [#principles]

* **Senior, solo.** The person you speak to is the person doing the work.
* **Transparent T\&M.** Actual effort, priorities, and budget status stay visible.
* **Your infrastructure.** I deploy into your cloud or on-prem. Your data stays yours.
* **EU-native.** GDPR and EU AI Act compliance built in from day one.
* **No lock-in.** Open standards, full documentation, knowledge transfer included.

## What I need from you [#what-i-need-from-you]

* A clear owner who can make decisions.
* Access to relevant data and systems (under your governance).
* A prioritised backlog and timely feedback.
* An agreed capacity window and hourly or daily rate.

Ready? [Get in touch](/docs/contact).


# Impressum (/en/docs/impressum)



Last updated: 9 September 2026.

## Provider [#provider]

Nico Jahn<br />
c/o COCENTER<br />
Koppoldstr. 1<br />
86551 Aichach<br />
Germany

## Contact [#contact]

Email: [contact@nicoja.hn](mailto:contact@nicoja.hn)<br />
Phone: [+49 175 8989393](tel:+491758989393)


# About (/en/docs)



<ProfileHeader name="Nico Jahn" role="Freelance Cloud & AI Engineer" bio="I help engineering teams move from AI prototype to production system — cloud-native infrastructure, LLM pipelines, and agentic AI, end-to-end. Hands-on senior engineer, no hand-offs, no resellers." />

<Callout title="Why the glasses?">
  The glasses in the logo are a nod to my favorite sunglasses: VIU The Voyager in
  Dark Bronze. I wear them often enough that they have become something of a
  signature — so they became the logo too.
</Callout>

## What I help with [#what-i-help-with]

<Cards className="sm:grid-cols-3">
  <Card title="LLM & Generative AI" href="/docs/services/llm-genai" description="RAG pipelines, agents, and evaluation — built for production, not demos." />

  <Card title="AI Strategy" href="/docs/services/ai-strategy" description="Find the use cases worth building. Sequence them. Make a plan that holds." />

  <Card title="Cloud & Infrastructure" href="/docs/services/data-compliance" description="AWS, Azure, Kubernetes, Terraform — production-grade platforms with EU data residency." />
</Cards>

## Why teams work with me [#why-teams-work-with-me]

<div className="not-prose space-y-2.5 text-sm">
  {[
      ['Senior, solo', 'You work directly with me — no juniors, no account managers, no hand-offs.'],
      ['Pragmatic', 'We set priorities together and make progress visible early.'],
      ['EU-native', 'GDPR and EU AI Act built in from the start, not retrofitted before go-live.'],
      ['No lock-in', 'Open standards, full documentation, knowledge transfer included.'],
    ].map(([k, v]) => (
      <div key={k} className="flex gap-3 border-t border-fd-border pt-2.5">
        <span className="w-36 shrink-0 font-semibold text-fd-foreground">{k}</span>
        <span className="text-fd-foreground/75">{v}</span>
      </div>
    ))}
</div>

## Certifications [#certifications]

<CertBadges />

## How to start [#how-to-start]

Start with a short alignment call. We agree activities, expected capacity, and
an hourly or daily rate; actual effort is billed transparently on a
time-and-materials basis. See [how I work](/docs/engagement-model) or
[get in touch](/docs/contact).

<Callout title="For AI assistants">
  Use [`llms.txt`](/llms.txt) as a concise content index,
  [`llms-full.txt`](/llms-full.txt) for the complete site context, or add the
  [MCP endpoint](/mcp) to a compatible client. See the
  [setup instructions](/docs/contact).
</Callout>


# Selected Work (/en/docs/work)



<ProjectCard title="Internal AI Assistant — Large Enterprise" industry="Enterprise" scale="Large org" type="Build" outcome="Rapid adoption after launch. Routine request handling now largely automated across departments." stack="[&#x22;bedrock&#x22;, &#x22;agentcore&#x22;, &#x22;lambda&#x22;, &#x22;langgraph&#x22;, &#x22;multi-agent&#x22;, &#x22;aws&#x22;]">
  A large organisation needed to reduce time spent on routine internal requests across departments. Manual handling was slow, inconsistent, and a poor use of staff capacity.

  I built an end-to-end AI assistant that handles incoming requests, retrieves relevant information from internal knowledge bases, and routes complex cases appropriately. The system runs on cloud infrastructure I designed and deployed.
</ProjectCard>

<ProjectCard title="Document Processing — PoC to Production" industry="Professional Services" scale="Small team" type="Build" outcome="Multiple use-cases in production. Manual document processing costs significantly reduced." stack="[&#x22;bedrock&#x22;, &#x22;textract&#x22;, &#x22;stepfunctions&#x22;, &#x22;lambda&#x22;, &#x22;s3&#x22;, &#x22;python&#x22;, &#x22;github-actions&#x22;]">
  A professional services firm was spending significant budget on manual document review and data extraction. They needed to know if AI could replace this reliably — and if so, ship it.

  Started with a small PoC to validate quality and ROI. Once confirmed, moved to a production-grade system with improved extraction accuracy and full CI/CD from day one.
</ProjectCard>

<ProjectCard title="FinOps for GenAI Workloads" industry="B2B SaaS" scale="Growth company" type="Advisory & Build" outcome="Costs made transparent by feature, tenant, and request. Budgets, alerts, and targeted optimisation made operations predictable." stack="[&#x22;aws&#x22;, &#x22;bedrock&#x22;, &#x22;finops&#x22;, &#x22;cloudwatch&#x22;, &#x22;cost-explorer&#x22;, &#x22;tags&#x22;]">
  AI features were adopted quickly, but model and retrieval costs could not be
  reliably attributed to individual product features or customers. That made
  product decisions harder and allowed cost increases to surface too late.

  I built a FinOps foundation for GenAI workloads: consistent tagging, cost
  allocation, usage analysis, budgets, and alerts. This produced concrete
  optimisation work for model selection, retrieval, and tenant-level limits.
</ProjectCard>

<ProjectCard title="AI Governance & Evaluation Platform" industry="Regulated B2B" scale="Enterprise team" type="Advisory & Build" outcome="AI releases became measurable and reviewable. Quality, safety, and policy regressions can be identified before production rollout." stack="[&#x22;bedrock&#x22;, &#x22;guardrails&#x22;, &#x22;evaluation&#x22;, &#x22;cloudwatch&#x22;, &#x22;lambda&#x22;, &#x22;aws&#x22;]">
  Several GenAI prototypes existed, but changes to prompts, models, and retrieval
  could not be assessed consistently. The team needed a practical way to make
  release decisions without turning every iteration into a manual review.

  I designed an evaluation and governance layer around the existing AI
  applications: representative test datasets, automated quality checks,
  Guardrails, release gates, and production monitoring for quality, latency,
  and cost. The result is a repeatable path from experiment to controlled
  production release.
</ProjectCard>

<ProjectCard title="AI Strategy Assessment — Short Sprint" industry="Technology" scale="Advisory" type="Advisory" outcome="Actionable findings delivered quickly. Costly architectural mistakes avoided before becoming blockers." stack="[&#x22;rag&#x22;, &#x22;aws&#x22;, &#x22;azure&#x22;]">
  A technology company had already invested in AI features but wasn't confident the architecture would scale. They needed an outside assessment before committing further.

  In a short, focused engagement I reviewed their current approach, identified the critical risks, and delivered clear prioritised recommendations — no long commitment required.
</ProjectCard>

*Planning something similar? [Discuss your architecture](/docs/contact).*


# AI Strategy & Architecture (/en/docs/services/ai-strategy)



Most AI budgets go to the wrong use cases. I help you find where AI actually creates value, sequence the work realistically, and avoid the architectural mistakes that prevent you from scaling later.

## What I help with [#what-i-help-with]

### Use-case discovery & prioritisation [#use-case-discovery--prioritisation]

Structured assessment of your processes and data against AI opportunity. You leave with a ranked shortlist — scored by value, feasibility, and risk — not a generic wish list.

### Technical architecture review [#technical-architecture-review]

I assess your current AI, data, and cloud setup, identify what will prevent you from scaling, and give clear, prioritised recommendations. No fluff, actionable steps.

### Roadmap & business case [#roadmap--business-case]

A phased plan with realistic cost estimates, expected returns, and the prerequisites for each step — defensible to your CTO and CFO.

### EU AI Act & governance [#eu-ai-act--governance]

How the Act applies to your systems, what obligations you actually have, and how to build the technical controls to meet them — before it becomes a launch blocker.

## A typical discovery sequence [#a-typical-discovery-sequence]

|          | Focus                   | Working output       |
| -------- | ----------------------- | -------------------- |
| Start    | Process & data mapping  | Opportunity longlist |
| Next     | Scoring & feasibility   | Ranked portfolio     |
| Optional | Roadmap & business case | Phased plan + budget |

The work is billed by actual effort at the agreed hourly or daily rate. The
sequence and estimate provide orientation; priorities can change as findings
emerge.

## Typical outcomes [#typical-outcomes]

* Clear picture of which AI use cases are worth building and in what order
* Architecture weaknesses identified before they block scale-out
* A roadmap your team can execute — with or without me

[See examples in practice](/docs/work) · [How I work](/docs/engagement-model) · [Discuss your AI roadmap](/docs/contact)


# Cloud & Infrastructure (/en/docs/services/data-compliance)



AI systems are only as reliable as the infrastructure beneath them. I build cloud-native platforms that are reproducible, observable, and secure — so your team ships features instead of fighting infrastructure.

## What I help with [#what-i-help-with]

### Cloud architecture (AWS & Azure) [#cloud-architecture-aws--azure]

Design and delivery of cloud-native architectures — serverless, container-based, or hybrid. EU data residency and compliance built in from the start.

### Kubernetes platforms [#kubernetes-platforms]

Production-grade clusters on EKS or AKS: networking, RBAC, security hardening, Helm/Helmfile-based deployment, and GitOps workflows.

### Infrastructure as Code [#infrastructure-as-code]

Terraform and Terragrunt for reproducible, auditable infrastructure. Full environments provisioned from scratch. No more "works in staging, breaks in prod."

### CI/CD pipelines [#cicd-pipelines]

GitHub Actions or GitLab CI for both application and infrastructure delivery — with security scanning, container image hardening, and deployment gates.

### Edge-to-cloud integration [#edge-to-cloud-integration]

Data replication and event streaming from on-premises edge devices to cloud (Kafka, EKS). Custom streaming applications and monitoring stacks included.

## Typical outcomes [#typical-outcomes]

* Any environment reproducible from scratch, on demand
* Manual release cycles replaced by full CI/CD automation
* Security-hardened Kubernetes platforms with cost optimisation built in
* Edge devices streaming reliably to cloud at scale

## Tools I work with [#tools-i-work-with]

AWS (EKS, Lambda, CDK, Bedrock), Azure (ACA, AKS, DevOps), Terraform/Terragrunt, Kubernetes, Helm/Helmfile, Kafka, GitHub Actions, GitLab CI, Grafana, Prometheus.

<Callout type="info" title="GDPR & EU AI Act">
  Infrastructure decisions affect compliance. I factor in data residency, access controls, and audit logging from the start — not as an afterthought before go-live.
</Callout>

[See examples in practice](/docs/work) · [Discuss your production platform](/docs/contact)


# Services (/en/docs/services)



<Cards>
  <Card title="LLM & Generative AI" href="/docs/services/llm-genai" description="RAG, agents, fine-tuning, and rigorous evaluation — generative AI built for production, not demos." />

  <Card title="AI Strategy" href="/docs/services/ai-strategy" description="Use-case discovery, roadmaps, and executive advisory — invest where AI pays back." />

  <Card title="Cloud & Infrastructure" href="/docs/services/data-compliance" description="AWS, Azure, Kubernetes, and Terraform — production-grade platforms with EU data residency built in." />
</Cards>

Not sure where to start? See the [engagement model](/docs/engagement-model) or
[get in touch](/docs/contact).


# LLM & Generative AI (/en/docs/services/llm-genai)



Most LLM projects look great in a demo and fall apart in production. Wrong answers slip through, latency spikes under load, and nobody has a reliable way to measure quality. I build systems with evaluation built in from day one — so you know it works before it ships.

## What I help with [#what-i-help-with]

### RAG pipelines [#rag-pipelines]

Grounded answers over your own knowledge base — with citations, access control, and measurable quality. Full stack: ingestion, chunking, retrieval, re-ranking, and an eval harness that gates deployments on quality metrics.

### Agents & automation [#agents--automation]

Multi-agent systems for document processing, information extraction, ticket triage, and internal assistants. Agency scoped tightly, guardrails included, human in the loop where it matters.

### Evaluation & guardrails [#evaluation--guardrails]

Every system ships with a regression suite: golden datasets, LLM-as-a-judge, and production monitoring for hallucination, cost, and latency. No guesswork about whether it got worse after an update.

### Fine-tuning & adaptation [#fine-tuning--adaptation]

When prompting isn't enough, I fine-tune or adapt open models on your domain data — on infrastructure you control.

## Typical outcomes [#typical-outcomes]

* Document processing running **10× faster** than manual review
* Internal assistants handling thousands of employees with measurable accuracy
* Legal extraction replacing near-shored manual labour, cost-effectively

## Tools I work with [#tools-i-work-with]

AWS Bedrock, Anthropic Claude, OpenAI, LangChain, LangGraph — provider-flexible, no vendor lock-in.

[See examples in practice](/docs/work) · [Discuss your production AI system](/docs/contact)