Apheris

🚀 Exciting to see our Series A funding milestone gaining traction! EU-Startups has spotlighted our Series A round, highlighting how Apheris is addressing one of the life sciences industry's toughest challenges: enabling secure, governed access to proprietary data for ML. With the backing of deep tech investors OTB Ventures and eCAPITAL ENTREPRENEURIAL PARTNERS, as well as health tech investor Heal Capital and Octopus Ventures we're empowering the life sciences industry with secure federated computing —empowering ML models with proprietary data while keeping it protected. Grateful for the recognition and excited for what's ahead! Dive into the full story: https://hubs.li/Q0313h6q0 #DeepTech #LifeSciences #AI #FederatedComputing #FundingNews

We’re thrilled to announce our Series A funding round – which was covered by TechCrunch 🔗 https://hubs.li/Q030ZHZK0 This milestone fuels our mission to power the largest and most secure life sciences data networks, tackling one of the most critical challenges of the industry: missing access to proprietary data. A heartfelt thank you to our new and existing investors: OTB Ventures, eCAPITAL ENTREPRENEURIAL PARTNERS, Octopus Ventures, Heal Capital, LocalGlobe, Dig Ventures as well as our incredible Angel Investors. We are grateful for your trust in us and your ongoing support. We’re excited to partner with you as we take Apheris to the next stage of growth! This achievement would not have been possible without the powerful dedication of the Apheris team and the trust of our customers – let's continue to drive forward collaboratively. #SeriesAFunding #FederatedLearning #LifeSciences #AI #DataCollaboration

🎄 Merry Christmas and Happy Holidays from all of us at Apheris! A big shoutout to our incredible employees, trusted investors, and visionary customers for making 2024 a year to remember. Your support, collaboration, and trust drive everything we do. 🎁 Stay tuned as we kick off 2025 with some exciting announcements that we can't wait to share! Wishing you a joyous holiday season and a fantastic start to the new year! #LifeSciencesInnovation #DataCollaboration #FederatedComputing

We’re hiring a Director of Commercial Operations - Life Sciences Apheris serves the world's leading Pharmas, Biotechs and TechBios. We’re tackling an industry wide challenge: publicly available data isn't large and diverse enough. Only through secure access to proprietary training data, the quality of AI models will improve such that they become applicable to industrial grade drug discovery & development research. As our Director of Commercial Operations, you’ll generate and advance new business opportunities in the life sciences industry. This role requires outstanding operational skills, solid technical and scientific understanding as well as domain expertise in dealing with complex Pharma and TechBio organizations. This is a unique role with highest impact on our next stage of growth, delivering industry-wide impact as orders of magnitude more training data can be leveraged within the life sciences networks we power. Get in touch if you have experience selling computational methods or models (particularly in drug discovery or early development), and excel at engaging with stakeholders across R&D and AI teams. 🔗 https://hubs.li/Q02_d3800 #Hiring #LifeScienceAI #PharmaAI #DataNetworks

Apheris has been featured in b2venture's TechBio market map at the foundational infrastructure layer. This recognition highlights the importance of secure, federated computing in life sciences to tackle one of the industry's biggest challenges: Secure access to proprietary life sciences data for AI model training. Key takeaways from the TechBio blog: 💽 Fragmented data The biopharma space is grappling with integrating massive fragmented datasets across organizations. The lack of interoperability and trust inhibits collaboration. Companies must adopt infrastructure solutions to securely connect and analyze these datasets without exposing sensitive intellectual property. 🛡️Trust in AI Transparency and traceability in AI workflows are critical for adoption in life sciences. Stakeholders want results and a clear understanding of how models arrive at their conclusions. Trustworthy infrastructure is key to enabling this level of transparency while ensuring compliance with rigorous regulations. 🤝Collaboration as a growth driver Collaboration among organizations is crucial for accelerating drug discovery and development. The ability to securely connect data and work together while protecting proprietary assets will shape the next generation of breakthroughs in precision medicine and beyond. 🔗 https://hubs.li/Q02-TZ-10 Thanks, Marisa Yasmin Krummrich. It's great to see Apheris as part of this dynamic and rapidly evolving ecosystem! 🔎 About us: Apheris enables secure data networks in healthcare and life sciences by safely connecting distributed data for analytics and AI. Our product, the Apheris Compute Gateway, is a federated computing infrastructure with governance, security and privacy controls. Data custodians can add their data to networks while staying in full control. Data scientists can access a larger data cohort and build better models. #TechBio #AIinLifeSciences #Federation

Federated data networks are becoming popular for many computer-aided drug discovery use cases. Interconnected nodes that are operationally independent, yet centrally accessible, allow researchers to train models that support their expertise in target discovery, prediction of molecular structures, or small molecule properties. We all know that quality research depends on access to quality data, and that data privacy must be well protected. In our latest blog post, Inken Hagestedt discusses various aspects of federated data networks, challenges that need to be addressed, and compares the operational efficiency of Federated Data Networks to other approaches. Worth a read for all who want to build and join data networks. https://hubs.li/Q02-j-rQ0 #drugdiscovery #federatedlearning #federateddatanetworks #drugresearch

🌍 This week, the Apheris team stepped out of their natural habitat—working in front of computers across the globe—and came together in Berlin for our Q4 office days. It was a packed, rewarding week filled with: ↗️ Big-picture discussions: Mapping out next year’s company direction, product goals & tech priorities. 🎓 Apheris University: - Hands-on hackathon with Ian Hales, Britta Srivas and Christopher Woodward focused on data harmonization and analysis - Internal security training lead by our security masters Alejandro Ortuno, and Muhammad Ibtehaj Akhtar - The legendary drug discovery pub quiz hosted by Johannes Forster and his new partner-in-crime Jan Stücke 💡Lightning talks: Sharing knowledge and insights through team presentations, including: - Evelyn Trautmann on her research into Parameter Efficient Fine-Tuning in federated learning scenarios. - Falko Krause’s deep dive into the UX of our governance portal. - Bojana Ivanovic and Eero Jyske’s session on key people and culture topics. Now we’re heading home—tired but fulfilled. These meetups remind us that Apheris is more than just a workplace—it’s a second home built on collaboration, connection, and shared purpose. Already looking forward to the next one! Shoutout to Ulrike Rahn for the amazing organization 👏 #Teamwork #RemoteFirst #OfficeDays

Robin Röhm, CEO & Co-Founder of Apheris, will speak at Bio-IT World in Boston, alongside John Karanicolas, Head of Computational Drug Discovery, AbbVie. Talk: "Transforming AI for Drug Discovery with Federated Learning and Proprietary Molecular Data" 🔎 The challenge: molecular data in the public domain is not diverse enough ML models don't reach the predictive accuracy needed for drug discovery, leaving a gap between algorithmic potential and real-world industrial needs. Today's ML models are mostly trained on the same publicly available datasets. These are commonly acknowledged as insufficient for the precision required for industrial use. 👉 The solution: Secure, collaborative AI model training on proprietary data Robin and John will explore how secure federated learning enables the collaborative training of ML models on proprietary datasets. They showcase how this approach can increase ML models' predictive accuracy and applicability domain while preserving data confidentiality and IP protection. 📆 When: April 3-4, 2025 📍Where: Bio-IT World Conference, Boston (Omni Boston Hotel) Thanks Bridget Kotelly for having us Join us to discover how secure AI collaboration unlocks new possibilities for industrial AI in drug discovery. #BioITWorld #BioITExpo #DrugDiscovery #FederatedLearning

While the northern hemisphere cools down, we’re turning up the heat with Apheris 3.4 - empowering the most secure federated data networks for the life sciences industry. Here’s the TL;DR of what’s new in our product: ✅ Reduced cognitive load for Data Scientists: Idle Compute Specs now deactivate themselves automatically, saving time and reducing resource usage. ✅ Close to a self-serve Model Registry: Independently update your models in the Model Registry. More independence = faster workflows! ✅ Enhanced Insights Into Active computations: Easily view details about active computations and their status. ✅ Markdown-Supported Dataset Descriptions: Create rich, detailed dataset descriptions with markdown, offering critical context to Data Scientists while keeping raw data protected. ✅ Streamlined CLI Experience: Faster logins and improved command performance. 🔎 Spotlight on the Apheris Model Registry At Apheris, we strive to make collaboration as easy and secure as possible. With Apheris 3.4, we're making it easier for data scientists to iterate on already approved models. Now, teams can adapt and improve their models to securely meet the dynamic realities of daily workflows. This enhancement ensures faster progress while maintaining the highest standards of security, privacy, and compliance that are essential for data collaboration. Are you ready for more details? Read Jan Stücke's full release blog: https://hubs.li/Q02Z8pps0

AI models for protein complex prediction don't generalize well Models, such as AlphaFold3, can predict the structure of protein complexes (e.g., protein-small-molecule interactions or protein-protein interactions). However, their applicability domain is often limited, especially for industrially relevant tasks. The challenge: Not enough data diversity in the public domain This limitation comes from the reliance on publicly available datasets to train these models. Only a fraction of the potential training data sits in the public domain. Much larger and more diverse data remains in silos - in private data repositories. The data is inaccessible due to IP concerns. Without access to more diverse training data, these models don't generalize well. The solution: Secure data networks for collaborative model training Secure data networks offer a solution to this problem, allowing access to private datasets while preserving the IP of the data. Owners retain physical and operational data control, allowing them to safely contribute their data to model training. All parties in a network benefit from a model with higher applicability for industrially relevant tasks. An example of this is the AI Structural Biology (AISB) Consortium (link as comment) #AI #DrugDiscovery #ProteinEngineering #MachineLearning