IKTOS

Their core work

IKTOS is a Parisian AI startup specializing in generative artificial intelligence for small molecule drug design. Their core product applies deep learning to predict molecular activity, ADME properties, and to generate novel drug candidates computationally — tasks that traditionally require years of wet-lab iteration. In MELLODDY, they contributed their predictive modelling expertise to a federated learning platform that let pharmaceutical companies train shared AI models without exposing proprietary compound data. In CARE, they applied those same computational tools to the urgent problem of identifying repurposed drugs against SARS-CoV-2.

What they specialise in

How they've shifted over time

IKTOS entered H2020 in 2019 focused on building the foundational infrastructure for privacy-safe, federated drug discovery AI — the emphasis on federated learning, distributed ledger, container orchestration, and predictive models in MELLODDY reflects a company helping establish the technical platform layer. By 2020, their second project pivoted that same AI toolkit toward an urgent real-world application: identifying repurposed drugs against COVID-19, with SARS-CoV-2 and repurposed drugs becoming the defining keywords. The trajectory is a clear move from infrastructure-building toward applied therapeutic discovery.

IKTOS is moving toward applied computational drug discovery — using their AI platform not just to build shared infrastructure but to generate concrete therapeutic leads, which positions them as a computational partner for both pharma R&D and future pandemic-preparedness consortia.

How they like to work

IKTOS has participated exclusively as a consortium partner and has never coordinated an H2020 project, suggesting they engage as a specialist contributor rather than a project driver. Despite only two projects, they have accumulated 51 unique consortium partners across 13 countries, which points to large, multi-stakeholder consortia — both MELLODDY and CARE are flagship collaborative projects with many industrial and academic members. This indicates they are comfortable operating inside complex, multi-partner structures where they deliver a specific AI/computational module rather than managing the whole.

IKTOS has built a surprisingly wide network for an organisation with only two projects — 51 unique partners across 13 countries, reflecting membership in two large, high-profile pan-European consortia. Their network is likely dominated by major pharmaceutical companies, academic research hospitals, and biotech firms given the nature of MELLODDY and CARE.

What sets them apart

IKTOS occupies a rare niche as an SME that brings production-grade generative AI for molecular design into EU research consortia — a capability more typically found inside large pharma informatics teams. Their dual expertise in federated learning (for multi-party pharma collaboration) and generative molecular AI means they can serve both as a privacy infrastructure specialist and as a drug candidate generator in the same consortium. For partners looking to add computational drug design without hiring a full in-house cheminformatics team, IKTOS is a compact, specialist entry point.

Highlights from their portfolio

• MELLODDY
  One of the largest federated learning initiatives in pharma history, involving ten major pharmaceutical companies sharing compound data under strict privacy constraints — IKTOS's participation signals they were trusted with highly sensitive industry data.
• CARE
  A flagship EU COVID-19 emergency response project (2020–2025, €421,638 to IKTOS), applying computational repurposing AI to one of the highest-priority public health problems of the decade.