Airy

Inspiration

Airy was inspired by the growing challenge of using AI in healthcare while protecting sensitive patient data. During the Brew Berlin Builder Program, I attended workshops on Zero-Knowledge Proofs and privacy-preserving technologies, which sparked my interest in how cryptography could enable trustworthy AI systems.

I started exploring how European hospitals could collaborate on machine learning without exposing confidential medical records or violating GDPR regulations.

---

What it does

Airy is a privacy-preserving AI platform that enables hospitals to collaboratively train AI models on encrypted patient data using Fully Homomorphic Encryption (FHE).

The platform allows healthcare institutions to improve AI diagnostics and predictions while keeping all sensitive patient information encrypted and stored locally.

No raw data is shared between hospitals.

---

How I built it

I explored concepts around federated learning, encrypted computation, and Fully Homomorphic Encryption to design a collaborative AI workflow for healthcare environments.

The system architecture focuses on:

• local encrypted data processing
• secure model collaboration
• privacy-first AI infrastructure
• GDPR-compliant workflows

I also researched how AI systems for conditions like sepsis could benefit from larger collaborative datasets across institutions.

---

Challenges I ran into

One of the biggest challenges was understanding which privacy-preserving technologies were practical for real-world AI systems.

Initially, I explored Zero-Knowledge Proofs, but I learned that FHE and federated learning were more applicable for collaborative machine learning workflows.

Another challenge was balancing privacy, scalability, and computational performance, since encrypted AI computation can still be resource intensive.

---

Accomplishments that I'm proud of

• Designing a clear privacy-first AI concept for healthcare
• Translating complex cryptography into a real-world use case
• Building an early-stage vision for trustworthy AI infrastructure
• Creating a project focused on both innovation and ethical responsibility

---

What I learned

I learned that privacy is becoming one of the biggest blockers for enterprise AI adoption.

I also learned that successful AI systems require not only strong models, but also trust, compliance, and secure collaboration mechanisms.

The project deepened my understanding of FHE, federated learning, healthcare AI, and the challenges of deploying responsible AI systems in Europe.

---

What's next for Airy

Next, I want to:

• validate the strongest healthcare use cases
• explore technical MVP development
• connect with engineers and healthcare partners
• improve scalability and performance
• further research privacy-preserving AI infrastructure for regulated industries

My long-term vision is to enable trustworthy AI collaboration across Europe without compromising individual privacy.

Built With

• designing-the-product-logic
• fhe
• my-focus-was-on-validating-the-concept