F.M. Delgado-Chaves

AI engineer · Agentic systemsOpen to Applied AI and forward-deployed roles, EU

Fernando M.Delgado-Chaves

I build agentic AI systems that do real science, grounded in nine years of computational biology, measured against public benchmarks, and inspectable at every step.

Most recentCo-Founder & CTO · Galagos
Record443 citations · h-index 8
DoctoratePhD · summa cum laude
Portrait of Fernando M. Delgado-Chaves

02 · The gap

Industrial-scale data generation, artisanal-scale analysis. Teams are not slow, they are overwhelmed. I build the agents that take that queue on.
Bioinformatician and scientist build: lab coat, microscope, teal molecular network. AI engineer build: navy circuit-print outfit, laptop, server stack and a branching blue agent structure. Founder build: amber and graphite engine with gears, prototype desk and a metrics display. Communicator and mentor build: microphone, teaching screen and violet radiating signal arcs.

03 · The bioinformatician & scientist

Signal, pulled from complexity.

From 2022 to 2025 I led network-medicine drug-repurposing initiatives inside REPO4EU, the EU consortium (grant No. 101057619), with an emphasis on infectious disease, building the multi-omics pipelines that turn disease modules into drug candidates, publishing with groups at TU Munich and Vienna, and with Harvard collaborators over the wider arc of the work.

Former Junior Group Leader for Generative AI for Medicine at the University of Hamburg, on a summa cum laude PhD and three master’s degrees. I review for Scientific Reports, PeerJ and the Journal of the Royal Society Interface, and have served as external examiner on three doctoral committees: Madrid, Granada and Seville.

Network medicine Multi-omics Drug repurposing Peer reviewer

04 · The AI engineer

Agentic systems that do real science.

I built my first conversational agent for disease-module discovery and drug repurposing in 2022, the direct precursor of ChatDRex, before “agentic” was a category. As CTO I led the architecture that grew out of it: model-agnostic multi-agent orchestration, a library of 250+ reusable agent skills, retrieval grounded in PubMed, UniProt and NCBI, and the evaluation harness we measured all of it against. My fix for tool-message streaming in ChatAnthropicVertex is merged upstream in langchain-ai/langchain-google (PR #1263).

LLM agents LangGraph RAG & hallucination control LLM evaluation

05 · The founder

Ownership of the whole engine.

Unusually for a CTO, I also ran the commercial engine: founder-led sales, marketing, and co-owning the EXIST application behind the approximately €234K in non-dilutive public funding the company raised. The team was selected into Antler’s inaugural Continental Europe residency in Berlin, roughly 100 founders from about 10,000 applicants, where the idea was pressure-tested hard. Antler did not invest and we did not raise.

≈ €234K non-dilutive Founder-led sales Antler residency

06 · The communicator & mentor

Making complex science legible.

I sit on the board of IABioMed, the Spanish Society for AI in Biomedicine, as Head of Communications. I taught a 12-session graduate course in systems medicine at the University of Hamburg, have supervised or mentored 19 students, and in 2026 took my seminar Your Research, Supercharged to the Max Planck Institute, CSSB Hamburg, the University of Porto and IABioMed, in English and Spanish.

IABioMed board 19 students supervised or mentored EN · ES · DE (basic)

07 · Selected work

Products & tools that shipped.

Published algorithms, open-source packages and shipped product work, each with a link you can check.

01

Galagos Co-Founder · CTO

An agentic “virtual bioinformatician” I carried from prototype to benchmarked release as CTO. On the full 205-task BixBench set it solved 166 of 205, 81.0%, at roughly €0.87 a question, and 92.0% on the 50-question expert-verified subset (46 of 50). Both runs were scored in-house, with the methodology published openly. NVIDIA Inception member; approximately €234K in non-dilutive public grants.

2025 to 2026 ↗
02

ChatDRex Concept · Co-senior author

A conversational, multi-agent drug-repurposing engine curated by the REPO4EU consortium, built with the students and colleagues I worked with at CoSy.Bio, on the conversational agent I had prototyped in 2022. One of three co-senior authors; manuscript under review.

Preprint ↗
03

AcadeMate Author

AI for systematic-review literature screening: the applied companion to my first-author PNAS work on LLMs in evidence synthesis.

GitHub ↗
04

DRaCOoN First author

A novel algorithm for pathway-level differential co-expression analysis in transcriptomics. First author, BMC Bioinformatics (2025). Reference implementation on GitHub, and on PyPI as dracoonpy.

BMC Bioinf. ↗
05

EnGNet / pyEnGNet Algorithm creator · contributor

Ensemble gene co-expression network reconstruction, the algorithm behind my PhD, originally in Java. Later reimplemented in Python and multi-GPU accelerated, published in The Journal of Supercomputing (2024).

pyEnGNet ↗

08 · Expertise

Three technical depths.

01

Agentic AI

LLM scientific agents; LangChain and LangGraph multi-agent systems; RAG, context engineering and hallucination control; LLM evaluation methodology, published openly.

02

Computational biology

Network reconstruction; differential co-expression; drug repurposing; multi-omics and biclustering.

03

Production systems

Python, R, Java and CUDA; multi-GPU and Nextflow HPC pipelines; model-agnostic architecture on single-tenant, GDPR-resident EU infrastructure.

09 · Research

Peer-reviewed, and checkable.

Google Scholar · verified 7 August 2026

443Citations
8h-index
11Peer-reviewed papers
6First-author papers

PNAS · 2025

Transforming literature screening: the emerging role of large language models in systematic reviews

First author · approximately 117 citations

Nature · 2025

Pathology-oriented multiplexing enables integrative disease mapping

Contributing co-author

BMC Bioinformatics · 2025

DRaCOon: a novel algorithm for pathway-level differential co-expression analysis in transcriptomics

First author

Artificial Intelligence in Medicine · 2019

Computational methods for Gene Regulatory Networks reconstruction and analysis: A review

First author · most-cited · approximately 217 citations

Nucleic Acids Research · 2021

Role of a cryptic tRNA gene operon in survival under translational stress

Co-author

10 · Recognition

Awards, talks & a public point of view.

2023

Young Investigator Award

REPO4EU

2023

Best Talk · RExPO’23

Stockholm

2022

Top 3 Speaker · RExPO’22

Maastricht

2026

Deep Tech Momentum · DTM100

Galagos · Berlin

2026

Falling Walls Venture

Nominated, April 2026

Series

“Your Research, Supercharged” · solo seminar

MPI · CSSB

A tall LEGO tower assembling from its parts: a microscope and molecule at the base,
                  a DNA helix and server racks in the middle, gears, books and a microphone above,
                  and a rocket at the summit.

Scattered parts

11 · Contact

Looking for the next system to make legible.