Bioinformatics AI Expert

Overview

Join a global AI community as a Bioinformatics AI Expert focused on advanced benchmark dataset creation and AI model evaluation. The work centers on designing reasoning-heavy, information-reversal tasks that test agent-style LLMs on real biological data from wet-lab and dry-lab workflows, including formats such as RNA-seq, FASTQ, BAM, and mzML.

What You'll Do

• Create complex scientific scenarios where models must infer hidden identities, experimental conditions, variant locations, or functional pathways from anonymized biological datasets.
• Package clean, self-contained workspace archives with real or synthesized datasets such as expression tables, sequence files, and metadata.
• Design tasks that require multi-step external tool use, including tools like BLAST, genomic interval utilities, variant annotators, phylogenetic software, and differential expression pipelines.
• Write step-by-step solution paths and strict grading rubrics with canonical answers and accepted variants for automated evaluation.
• Verify benchmark deliverables for leakage, determinism, and automated validation compatibility.

Requirements

• Academic background or degree in Bioinformatics, Computational Biology, Genomics, Systems Biology, Molecular Biology, or Biomedical Data Science.
• At least 3 years of hands-on bioinformatics experience with real biological data types such as WGS/WES, RNA-seq, single-cell, ChIP/ATAC-seq, mass spectrometry, or metagenomics.
• Practical experience with standard bioinformatics command-line tools and software suites such as samtools, bedtools, BLAST, DESeq2, snpEff, Bioconductor, and GATK.
• Proficiency in Python or R for data anonymization, filtering, and workspace packaging.
• Strong understanding of benchmark validation, prompt engineering, RLHF, or LLM agent evaluations.
• C1+ or native English proficiency for precise prompt writing, clear documentation, and rubric authoring.

Assessment & Onboarding

Successful applicants will complete a Bioinformatics Technical Sandbox Assessment by creating one sample information-reversal task, including the dataset, rubrics, and chain-of-thought, and will also need to complete identity verification before joining the active project pool.