I help companies build AI systems that actually ship.

Head of AI at ISTEC · PhD in Biomedical Engineering · Published in Nature Communications

Most AI projects stall between prototype and production. I bridge that gap. With a research background in computational oncology and hands-on experience shipping ML systems, I help organizations move from experiments to real business value.


How I Help

Three ways I help companies ship AI that works

From Prototype to Production

ML Pipeline Development

Your data science team built a promising model, but it's stuck in notebooks. I design and build production ML pipelines that take your models from experiment to deployment -- reproducible, scalable systems your team can maintain.

Models in production within weeks, not months
Automated training, testing, and deployment pipelines
Team upskilled to maintain the system independently

For companies with data and models but no production ML infrastructure

Better Models, Faster

AI Model Training & Optimization

Your model works but accuracy isn't where it needs to be, or training takes too long. I optimize architectures, tune hyperparameters, and redesign data pipelines to get measurably better results from your existing data.

Measurable accuracy improvements with clear benchmarks
Reduced training time and inference costs
Data preprocessing and feature engineering that compounds

For teams with working models that need to perform better

Scale Without Breaking

Cloud AI Infrastructure

Your AI workloads are growing but your infrastructure isn't keeping up. I architect cloud-native ML platforms with distributed training, automated MLOps, and auto-scaling that grows with your needs.

Cloud costs optimized without sacrificing performance
CI/CD for ML: automated testing, versioning, deployment
Infrastructure that scales up and down automatically

For organizations scaling AI beyond a single machine

Know Where to Start

AI Strategy & Roadmap

You know AI could add value but aren't sure where to begin or what's realistic. I assess your data, team, and goals to build an honest roadmap: what to build first, what to skip, and what return to expect.

Clear assessment of what AI can and can't do for your business
Prioritized roadmap with quick wins and long-term bets
Team and infrastructure requirements mapped out

For companies exploring AI for the first time


Nature Communications

Published Research

6+

Peer-Reviewed Publications

PhD

Oregon Health & Science University

Head of AI

at ISTEC

Products I've Shipped

Real results from real systems

85%

Scheduling Time Reduction

99%

Fewer Scheduling Conflicts

Built and shipped at

ISTEC · Oregon Health & Science University · Anobrain AI · Origin Audio

Things I've Built

Products and platforms I've designed, built, and shipped

Schedule Genius

SaaS Platform Live Product

85% reduction in scheduling time for university admissions. Intelligent recruitment event management platform with smart matching algorithms and automated coordination. Trusted by leading graduate programs.

85% Time Reduction
99% Fewer Conflicts
4.9/5 Admin Satisfaction
Optimization Algorithms Smart Matching Real-time Dashboard Academic Systems

Vector Database Explorer

Research Platform Live Platform

Turned published research into a usable product. Exploration platform for multimodal cancer data through vector embeddings, with 3D interactive visualization and high-performance similarity search. Built on findings from peer-reviewed publications.

Multimodal cancer data analysis through vector embeddings
3D interactive UMAP visualizations of embedding spaces
ChromaDB integration with comprehensive metadata support
Vector Embeddings ChromaDB UMAP Visualization Biomedical Research

PhD by AnoBrain.ai

SaaS Platform Live Product

AI-powered platform for academic researchers. Streamlines research project management with AI journal suggestions, milestone creation, and smart calendar integration. Built for students, postdocs, and research labs.

AI-powered journal suggestions and milestone creation
Smart calendar integration and deadline tracking
Institutional licenses for labs and universities
AI/ML Academic Research SaaS Development Calendar Integration

Origin Audio

Web Service Live Service

AI audio processing for musicians, no signup required. Platform with intelligent auto-slicing, stem isolation, batch normalization, and format conversion. Free tier up to 25MB with no account needed.

AI Auto Slicer for automatic instrument sample organization
Stem Isolator for vocals, drums, bass, and instruments
Batch normalization and file conversion with unlimited processing
AI Audio Processing Stem Separation Batch Operations Cloud Service

Divey

Mobile Application Live Product

AI-powered dive safety on iOS and Android. Buoyancy calculator app using advanced algorithms for precise calculations, making diving safer and more convenient.

Available on iOS and Android platforms
AI-powered safety calculations for diving
AI/ML Mobile Development Safety Systems Cross-Platform

Professional Experience

Building the future through innovation and expertise

Head of AI

ISTEC · Karlsruhe, Germany

Leading AI strategy, research, and implementation at ISTEC, driving innovation and transforming business processes through cutting-edge artificial intelligence solutions and intelligent system architectures.

Technical Expertise
AI Strategy Machine Learning Deep Learning Team Leadership Data Science
February 2026 - Present
Active

Founder & AI Strategy Director

Leading strategic AI consulting firm, transforming businesses through data science excellence and next-generation intelligent solutions. Delivering comprehensive strategies and training for organizational advancement.

Strategic Consulting
AI Strategy Data Science Custom Training Business Intelligence Digital Transformation
Business Value

Proven success in executing marketing strategies and delivering tailored training for improved client operations and data-driven business evolution.

May 2022 - Present
Active

Senior AI Software Engineer

Lead developer of EmbKit, a modular library for composing VAEs and other generative models to learn embedding spaces under explicit constraints. Ships with out-of-the-box templates and a flexible API to customize architectures, priors, losses, and evaluation. Designed to cut prototyping time and make constraint-driven representation learning practical.

Technical Expertise
Variational Autoencoders Generative Models Representation Learning Library Design Python
September 2025 - December 2025
Completed

AI Consultant & Technical Mentor

Accelerating client success through personalized AI and machine learning mentorship, delivering cutting-edge technical guidance across diverse industries and skill levels.

Teaching Expertise
Angular Development Python Programming Machine Learning Data Science Cloud Architecture
Client Success

Customized, one-on-one guidance to boost technical skills and confidence, helping clients meet specific project needs and accelerate learning goals.

August 2022 - December 2025
Completed

Founder & Technical Architect

Origin Audio LLC

Co-founded innovative AI-powered platform revolutionizing music production through intelligent audio processing and machine learning solutions. Led end-to-end product development and cloud infrastructure management.

Technical Leadership
Full-Stack Development Azure Cloud API Architecture ML Audio Processing DevOps
Innovation Impact

Experience in scalable AI application and service delivery, aligning technical solutions with creative industry needs and user-centered design principles.

January 2021 - December 2025
Completed

PhD Researcher in Computational Oncology

Pioneering the next generation of AI-driven cancer research through advanced machine learning architectures and computational biology innovations. Creating robust algorithms for data-driven insights in precision oncology.

Technical Expertise
Graph Neural Networks Variational Autoencoders Deep Learning Spatial Omics Predictive Analytics
Impact & Value

Expertise in creating machine learning models to uncover novel insights from complex datasets, advancing data-driven solutions in precision medicine, healthcare, and beyond.

September 2019 - August 2025
Completed

Published Research

Published in Nature Communications, npj Precision Oncology, Genome Biology, and Cancer Research

Predicting anti-PD-1 immune checkpoint blockade response in melanoma patients with spatially aware machine learning models

Raphael Kirchgaessner et al.
npj Precision Oncology, January 2026 Latest Publication

Applied single-cell spatial proteomics together with machine learning to predict advanced melanoma patient response to anti-PD-1 immune checkpoint blockade therapy. ML models integrating multiple molecular features accurately predicted response in 11 of 12 patients, uncovering key tumor microenvironment features driving response.

Spatial Proteomics Immune Checkpoint Melanoma Machine Learning Tumor Microenvironment Precision Oncology

Aggregating multimodal cancer data across unaligned embedding spaces maintains tumor of origin signal

Raphael Kirchgaessner, Kaya Keutler, Layaa Sivakumar, Xubo Song, Kyle Ellrott
Biorxiv, May 2025

Machine Learning algorithms that preserve the underlying signal when aggregating multimodal cancer data, advancing computational approaches for comprehensive cancer analysis.

Multimodal Data Embedding Spaces Cancer Analysis Signal Preservation Vector Database Oncology Research

Imputing Single-Cell Protein Abundance in Multiplex Tissue Imaging

Raphael Kirchgaessner, Cameron Watson, Allison Creason, Kaya Keutler, Jeremy Goecks
Nature Communications, May 2025 Published

Innovative approach to impute protein abundance in tissue imaging, advancing the analysis of complex biological systems and enhancing data interpretation in biomedical research.

Single-Cell Analysis Protein Abundance Tissue Imaging Biomedical Research Deep Learning Oncology Research

SyntheVAEiser: augmenting traditional machine learning methods with VAE-based gene expression sample generation

Brian Karlberg, Raphael Kirchgaessner, Jordan Lee, Matthew Peterkort, Liam Beckman, Jeremy Goecks, Kyle Ellrott
Genome Biology, December 2024 Published

SyntheVAEiser, a variational autoencoder-based tool, synthesizes gene expression samples to augment datasets for machine learning tasks. Trained and tested on over 8,000 cancer samples, enhancing prediction accuracy for cancer subtypes.

8,000+ cancer samples analyzed
Improved prediction accuracy for underrepresented cohorts
Variational Autoencoders Gene Expression Cancer Subtypes Synthetic Data Oncology Research

Tumor model to tumor treatment: Applying deep learning approaches to map multimodal data

Brian Karlberg, Raphael Kirchgaessner, Jeremy R. Jacobson, Kyle Ellrott, Sara J. Gosline
Cancer Research, March 2024 Published

Novel VAE-based strategy for correcting platform effects in cancer drug response data, enhancing cross-model prediction accuracy and advancing translatability of proteogenomic studies.

Deep Learning Drug Response Proteogenomics Cross-Model Prediction Oncology Research

Computational Pipeline to Identify Gene Signatures Defining Cancer Subtypes

Ekansh Mittal, Vatsal Parikh, Raphael Kirchgaessner
bioRxiv, November 2022

Developed a novel computational pipeline for identifying gene signatures in cancer subtypes, demonstrating the power of data-driven approaches in oncology research.

Computational Pipeline Gene Signatures Cancer Subtypes Oncology Research

Let's Talk About Your AI Challenge

Whether you have a clear project in mind or just want to explore what's possible, I'm happy to talk it through.

Book a Free 30-Minute Strategy Call

The fastest way to get started. We'll discuss your situation, I'll share honest feedback on what's feasible, and you'll walk away with actionable next steps.

or send a message directly

Send a Message

I'll get back to you within 24 hours

Not sure what you need yet? That's fine. Book a call above and we'll figure it out together.

Email

exitare@exitare.de

Location

Karlsruhe, Germany

Response Time

Within 24 hours