AI Biologist - Cancer Biology (Applications)

LatchBio; Deutschland; Vor Ort; EUR 103.000 - 155.000; Vollzeit

Benefits dieser Stelle

Free meals at office

Unlimited snacks

Team outings and events

Office in SF waterfront

Zusammenfassung

LatchBio is seeking an AI Biologist to develop benchmarks that translate raw molecular tumor data into concrete, defensible clinical decisions. You will design tasks, specify reference analyses, and create scoring criteria across cancer biology domains—ensuring robust handling of variability and data discordance.

The role requires hands-on cancer biology experience, strong Python/R skills, and the ability to articulate uncertainties and alternative interpretations in writing.

Qualifikationen

• Proficiency in Python and/or R for multi-omic data analysis.
• Experience with single-cell analysis (Scanpy/Seurat) and gene-set scoring (GSEA/ssGSEA).
• Understanding of cancer-genomics standards and interpretation workflows.

Aufgaben

• Define defensible reference analyses and scoring criteria for cancer biology benchmarks.
• Turn cancer datasets into rigorously graded tasks with deterministic scoring.
• Anticipate variability, data conflicts, and mechanism-driven tradeoffs in analyses.

Kenntnisse

Python

R

VCF/MAF parsing

Single-cell analysis

Statistical analysis

CIBERSORTx

Jobbeschreibung

AI Biologist - Cancer Biology (Applications)

Latch builds rigorous, scientific benchmarks (benchmarks.bio) for AI agents across biology. We work with frontier labs and pharma to measure whether agents can handle real biological workflows; the messy, multi-layer reasoning that matters in the field.

About the Role

Our work in Cancer Biology means measuring whether an agent can take a tumor case from raw molecular data to a real clinical decision. You’ll identify real cancer-biology datasets and turn them into rigorous, deterministically-graded tasks.

Our benchmarks are agentic and cross-domain. Each task hands an agent the artifacts a cancer researcher would actually have and asks for a concrete conclusion. Your job is to establish defensible reference analyses and scoring criteria across conclusion types, the benchmark measures that span basic cancer biology to clinical outcomes. You’ll anticipate where agents take plausible but wrong turns and test whether they can handle experimental variability, incomplete data, conflicting evidence, and mechanism-driven tradeoffs.

Our team emphasizes cross-domain integration and long-horizon reasoning.

Requirements

• 1+ years personally working in basic, translational, or clinical cancer biology. Experience with hands on analysis of multi-omic tumor data hands-on somatic/germline variant calling and interpretation (WES/WGS/panel; MAF/VCF), copy-number and structural-variant analysis, mutational signatures, bulk and single-cell RNA-seq (cell-state annotation, subtyping), tumor-immune contexture (immune deconvolution, spatial transcriptomics, multiplex imaging), subclonal reconstruction, or functional-genomics screens (CRISPR/RNAi), or clinical research is highly preferred.

• Proficiency in Python and/or R: VCF/MAF parsing, single-cell analysis (Scanpy/Seurat), gene-set scoring (GSEA/ssGSEA), immune deconvolution (CIBERSORTx), subclonal reconstruction (PyClone), statistical analysis.

• Understanding of cancer-genomics standards: AMP/ASCO/CAP and ACMG/AMP variant tiers, ESCAT and OncoKB actionability levels, PAM50 and Consensus Molecular Subtypes, IFN-γ/T-cell-inflamed signature, consensus Immunoscore.

• Recognize experimental variability, and assay limitations; distinguish real biological signal from pipeline or annotation artifact.

• Strong written communication on technical decisions, uncertainty, and alternative interpretations.

Nice-to-Have

• Familiarity with cancer-genomics resources and standards: TCGA/GDC, PCAWG, AACR Project GENIE, cBioPortal, OncoKB/CIViC/VICC, COSMIC signatures, HTAN, Human Cell Atlas, DepMap/Project Score, gnomAD/ClinVar, MSigDB.

• Breadth across multiple conclusion types like: diagnosis, mechanism, clonal evolution, immune contexture, target nomination, actionability.

• Prior work on benchmarks, task-based assessment, or deterministic grading.

Culture @ Latch

How we work. Genuinely flexible schedules - we just ask that you communicate when you’re coming in later than usual. We care most about hard work and output. We’re respectfully opinionated, it will always be us against problems, not each other. Optimize for each other’s time and bring solutions, not just problems. You’ll join a bench suited to your expertise, but we value cross-domain learning and interoperability across teams.

Office & Perks. Waterfront office near Oracle Park, 2x free daily meals, unlimited snacks, company outings most months, and team offsites. Plus a vibrant community: cycling, soccer, figure skating, boxing, run clubs, reading clubs, martial arts, music, game nights.

Compensation & Logistics

• 1099 (or W8-BEN) contract, 40 hrs/week, no end date

• Fully performance-based pay - OTE: $120K–$180K, uncapped upside. 2x quota = 2x pay.

• 2-4 week paid ramp: full OTE during ramp

• Remote (globally), hybrid, or onsite in SF (onsite preferred)

• Work authorization: OPT visa holders only (not STEM Extension)

• Onsite perks: 2x free meals/day, waterfront office (China Basin), monthly parking (based on availability)

• Candidates with all of the above qualifications + proven management skills will be eligible for more senior positions

Interview Process

Timeline: We move fast: 8 - 12 days from submission to offer.

1. Intro Screen - Technical Recruiter

2. Take-Home Project - HackerRank

3. Technical Interview - Member of Technical Staff

4. Culture Interview - C-Suite

5. Offer