data-driven ai & agents

Data. AI. Agents.
Results that count.

Data-driven, AI-enabled and agent-enabled solutions: AI agents that automate real work, LLM and RAG applications, machine learning and computer vision. Built from Switzerland for clients worldwide, in the cloud or fully private when your data must stay under your control.

Book a free 30-min consultation →See case studies
  • 40+ LLMs in production
  • Research on arXiv
  • Based in Switzerland · projects worldwide

What I Do

Services

AI Agents & Process Automation

Your team spends hours on repetitive work: reading, sorting, extracting and reporting.

AI agents that research, extract, classify and score, so people only make the final decision.

  • Multi-step LLM workflows and agent pipelines
  • Document and email triage, data extraction
  • LLM outputs embedded in your dashboards and tools

LLM APIs · Playwright · FastAPI · React

AI Knowledge Assistants (RAG)

Knowledge is buried in manuals, policies, tickets and reports, and nobody finds it in time.

An assistant that answers questions from your own documents, with sources your team can check.

  • Document ingestion, embeddings and semantic search
  • Answers with citations to the source documents
  • Automatic summaries of documents and system logs

Embedding models · Vector search · FastAPI · Python

Data Science, Machine Learning & Computer Vision

Your data and images hold answers, but there is no reliable way to analyse them at scale and act on them.

Production-ready models and dashboards that turn images and data into decisions.

  • Predictive models and data-driven decision support
  • Dashboards, ETL pipelines and analytics
  • Medical and industrial image analysis (2D/3D)
  • Scenario and digital-twin modelling

PyTorch · TensorFlow · Azure ML · SQL · Power BI · R Shiny

Private & Secure AI Infrastructure

You want ChatGPT-level productivity, but client and company data must stay under your control.

Your own LLM platform, on-premise or in the data-centre region you choose, with no external API dependency.

  • Selection and deployment of open-source LLMs
  • Secure internal chat interface for your team
  • Access control, logging and model updates

Ollama · Open WebUI · Llama · Mistral · Docker · Azure

Ways to Start

Small first steps, clear results

Half day · fixed price
AI Opportunity Workshop

We map your processes, identify the 2–3 AI use cases with the best return, and leave you with a prioritised roadmap.

4 weeks · fixed scope
AI Pilot: Agent, RAG or ML

A working prototype on your own data, tested with real users before you invest further.

1–2 days · fixed price
AI & Data Readiness Check

A review of your data, tools and privacy requirements (GDPR, nDSG), with clear recommendations on which AI approach fits: cloud, hybrid or on-premise.

Case Studies

Selected Work

Private AI

Private LLM Platform for a Public-Sector Organisation

Challenge
Staff needed generative AI for documents and IT operations, but sensitive data could not be sent to external AI providers.
Result
Generative AI in daily use, with no data leaving the organisation and no dependency on external APIs.
  • Ollama
  • Open WebUI
  • RAG
  • Embeddings
  • C#
  • Azure DevOps
Read case study →
Medical AI · Research

Self-Supervised 3D Cancer Cell Segmentation

Challenge
A medtech company needed to segment cancer cells in very large 3D light-sheet microscopy scans, where expert annotations are slow and expensive.
Result
Segmentation 3% better than expert-annotated masks and 20% higher downstream diagnostic accuracy. Published on arXiv and under review at Pattern Recognition (Elsevier).
  • PyTorch
  • Swin Transformer
  • Self-Supervised Learning
  • Azure ML
Read case study →
AI Agents

Agentic AI Pipeline for Research & Scoring

Challenge
Screening hundreds of online listings by hand, finding the original source and judging relevance took hours every week.
Result
A daily manual task turned into a ranked shortlist, ready for a human decision.
  • LLM APIs
  • Playwright
  • FastAPI
  • React
  • SQLite
Read case study →
Data Science · Modelling

Digital Twin for Population, Traffic & Energy

Challenge
Planners needed to understand how population growth and migration would affect traffic, energy use and CO₂ emissions across Bosnia and Herzegovina.
Result
CO₂-reduction and cost-optimisation scenarios for Sarajevo, based on one consistent, data-driven model.
  • Python
  • GIS
  • Scenario Modelling
  • Energy Systems
Read case study →

How I Work

From idea to production in four steps

  1. Discover

    A free 30-minute call, then a workshop to understand your processes, data and privacy requirements.

  2. Pilot

    A small, fixed-scope prototype on your real data, so you see the value before investing further.

  3. Deploy

    Production rollout in your cloud or on your own infrastructure, integrated with the tools your team already uses.

  4. Support

    Monitoring, model updates, training for your team and continuous improvement.

About

Who you will work with

Morteza Kiani, founder of Swiss Digital Science
Morteza Kiani
Founder · AI Engineer & Data Scientist
Risch-Rotkreuz, Switzerland · working worldwide

I’m an AI engineer and data scientist with a background in both data science and electrical engineering. I help companies in Switzerland and abroad turn data into decisions and automate work with AI and agents.

My experience spans running private LLM infrastructure in the Swiss public sector, self-supervised medical image segmentation for medtech, data pipelines and dashboards for diagnostics manufacturing, and digital-twin modelling with ETH Zürich. My research is published on arXiv (2509.06885) and is under review at Pattern Recognition (Elsevier).

Every engagement follows the same discipline: understand your domain, choose the simplest approach that works, build it so it runs reliably, and deliver results you can verify.

LinkedIn →GitHub →Publication: Barlow-Swin (arXiv) →
2×
MSc degrees: Data Science (HSLU) and Electrical Engineering (Politecnico di Milano)
40+
LLMs deployed and operated in production
arXiv
Research in medical image segmentation, under review at Pattern Recognition
4
Languages: English, German, Italian, Persian
Research-Grade Work
Methods from published research, applied to real business problems and measured honestly.
End-to-End Delivery
From the first workshop to a running system. One person who understands both the model and the infrastructure.
Privacy by Design
Data protection is planned in from the start, from cloud setups to fully private, on-premise AI.

FAQ

Common questions

Do you work with clients outside Switzerland?

Yes. I am based in Switzerland and work with clients across Europe and worldwide, remotely and, where useful, on-site. Projects run in English, German or Italian.

Where does our data go?

That is your decision. Depending on your requirements, solutions run with established cloud AI providers, in a data centre in the region you choose, or fully on-premise so that no data leaves your company. Privacy rules such as the GDPR and the Swiss nDSG are part of the design from the start.

Are open-source models good enough compared to ChatGPT?

For most business tasks, such as answering questions from documents, summarising, extracting data and drafting text, current open-source models like Llama and Mistral perform very well. In the pilot we test them on your real tasks, so you decide based on results.

How long does a project take?

A workshop takes half a day. A pilot on your own data usually takes about four weeks. A production rollout depends on your systems, typically one to three months.

What does it cost?

Entry offers have a fixed price, agreed up front. Larger projects are quoted after the workshop, once the scope is clear. The first 30-minute call is always free.

Do we need our own GPU server?

Not necessarily. Depending on model size and usage, options range from an existing server to a dedicated GPU machine or a cloud provider. I help you choose the most cost-effective setup.

Get in Touch

Book a free 30-minute consultation

Contact Details
Emailinfo@swissdigitalscience.ch
LocationRisch-Rotkreuz, Switzerland · working worldwide
LinkedInlinkedin.com/in/morteza-kiani
Tell me briefly about your processes or data. In a free 30-minute call we will look at where AI can help and what a first step could be. No obligation, response within 2 business days.