Bioinformatics / Computational Genomics Expert

What you'll do

You write the exam, not sit it. For GeneBench Pro you design challenging, agentic computational genomics tasks that an AI agent must solve by interpreting biological data, writing and running code and working across scientific files, and you build the reference solution and the grader.

Each task needs realistic data (FASTA/FASTQ, VCF, BAM/SAM, BED, TSV/CSV, annotations, expression data, germline or somatic variants), a clear specification, input datasets, an expected output schema and a deterministic or objectively verifiable ground truth. The ad stresses tasks that test scientific judgment: messy or ambiguous data, spotting artifacts, choosing the right analysis, revising assumptions as results come in, and knowing when a conclusion is decision-ready. You also validate that each task is scientifically correct, solvable from what is supplied and hard enough for frontier models, design grading that separates correct solutions from plausible-looking ones, and fold in reviewer feedback.

Who qualifies

• A PhD, postdoctoral experience or equivalent research experience in bioinformatics, computational biology, genomics, computational genetics or a close field
• Strong hands-on Python
• Experience analyzing biological sequence or genomics datasets
• Comfort in Linux and the command line

Preferred: genomics workflow experience (variant analysis, transcriptomics, phylogenetics, population or clinical genomics), tools such as Biopython, pandas, samtools, bcftools, BLAST or PLINK, reproducible pipelines and prior evaluation of AI on scientific tasks. Bonus: coding-agent experience, benchmark or grader design, relevant publications, Docker.

What it pays

$260 per approved task, as stated in the ad. "Approved" matters: pay depends on tasks passing review, so reviewer feedback directly affects what you earn. The ad does not say how many tasks to expect in the 5 weeks, so the total is not knowable from the listing.

Worth knowing

Good:

• A published per-task figure, rare among Turing listings
• Work that uses real domain expertise rather than generic rating
• Short, defined contract

Less good:

• At least 6 hours a day and a minimum of 40 hours a week, including a 4-hour overlap with PST, is a full-time load for 5 weeks
• Per-task pay with approval gating makes the hourly equivalent uncertain
• Contractor assignment with no medical or paid leave
• The start date is stated as "next week", so timing is tight

The ad describes no assessment. Turing's help centre mentions skill tests and, for some roles, a live interview.

About this listing

Read from Turing's job board on 18 October 2026. The pay, hours and duration are Turing's text; the interpretation is ours. See Turing.