OMICs Scientist

OMICs Scientist

Spore.Bio

Paris, France

About the role

This is not a supporting position. You will own the scientific interpretation of multi-omics data for our AMR platform, working from raw genomic, transcriptomic, and proteomic or metabolomic data through to biological insight that directly shapes what our models learn and how we validate what they find. Your expertise will guide model development and interpretation within a highly collaborative, multidisciplinary environment, working as a scientific peer to our deep learning team rather than a data preparation function upstream of it.

Main Responsibilities

• Multi-omics characterisation of resistance: You will work across genomics, transcriptomics, and at least one of proteomics or metabolomics to build a mechanistic picture of resistance, persistence, and tolerance in our strain library and clinical isolates. This includes bringing a working understanding of bacterial population structure, evolutionary dynamics, and how resistance spreads and is maintained, to bear on how we design experiments and interpret their output;
• Genotype-phenotype linkage: You will work with paired genomic and quantitative susceptibility data (AST, MIC) to characterise where sequence-based prediction holds and where it diverges from measured phenotype. You will bring resistome literacy at a mechanistic level, not just annotation-level familiarity, and will be the person in the room who can explain why a genotypic call and a phenotypic result disagree;
• Data curation for model-ready pipelines: You will interrogate, curate, and structure complex, heterogeneous, and often clinically-derived biological data (underrepresented sequences, mixed populations, ambiguous or noisy signal) so that it is usable for training pipelines without losing scientific rigour along the way. You will apply alignment-free and reference-free methods where standard approaches fall short, and you will know the failure modes of building generalisable models from real-world biological data well enough to flag them before they surface downstream;
• Working as a peer to the deep learning team: You will contribute directly to model design and biological framing, not only to data preparation. This means having informed views on feature representation, embedding strategies, and multi-omics data fusion for microbial systems, and being able to evaluate model outputs from a biological standpoint, catching what looks statistically sound but biologically implausible;
• Scientific contribution: You will contribute to publications, conferences, and intellectual property arising from this work, and collaborate closely with microbiologists, computer vision experts, and optical physicists to keep the biological interpretation of our models grounded.

About you

We are looking for someone who takes ownership, works with rigour in a fast-moving environment, and finds genuine satisfaction in building a scientific programme that others can rely on. We value curiosity, initiative, and a growth mindset, along with a strong critical awareness of what a given method can and cannot tell you.

• Academic background: A PhD in microbial genomics, computational microbiology, bioinformatics, or a closely related discipline is required;
• Multi-omics expertise: +10y of direct experience with genomics, transcriptomics, and at least one of proteomics or metabolomics in a microbial context is required. You should be comfortable moving between data layers and know which one to reach for when a question can't be resolved with sequence data alone;
• Mechanistic microbiology: A working understanding of bacterial population structure, evolutionary dynamics, and the mechanisms by which resistance spreads, not just how to run the pipeline that detects it;
• Genotype-phenotype fluency: Experience working with paired genomic and quantitative susceptibility data, with a clear-eyed view of where sequence-based inference is reliable and where it breaks down relative to measured phenotype;
• Data handling under real-world conditions: Demonstrated ability to work with heterogeneous, clinically-derived datasets, including underrepresented sequences, mixed populations, and noisy or ambiguous signal, and to make sound judgement calls about signal versus noise;
• Method versatility: Proficiency with alignment-free and reference-free approaches for cases where standard reference-based methods are insufficient;
• Collaboration with computational teams: Comfortable working as a scientific peer to deep learning scientists, with informed views on feature representation and multi-omics data fusion, and confident evaluating model outputs against biological plausibility;
• Communication: Clear written and verbal expression in English. Comfortable synthesising complex biological findings for non-specialist audiences and contributing to publications, patents, and partner-facing material.