Bioinformatics Engineer — Single-Cell AI
About the position
At LatchBio, our AI agents help thousands of scientists analyze and interpret data across the full stack of modern multi-omic technologies — starting with single-cell and spatial, and expanding fast.
We're building the ground truth for AI in single-cell biology. Our benchmark scBench — 394 verifiable problems across six sequencing platforms — shows the best frontier model today still fails nearly half the time. We're hiring bioinformatics engineers to close that gap: scientists who can turn real experimental data into the precise, falsifiable questions that define what it means for an AI agent to actually understand scRNA-seq.
Responsibilities
• Own end-to-end scRNA-seq analyses across multiple projects: raw platform outputs → QC and failure diagnosis → normalization → dimensionality reduction → clustering → cell typing → differential expression → trajectory analysis → defended biological claim.
• Build reproducible workflows and produce clear decision traces: what was filtered, why, what changed the conclusion, what would falsify the claim.
• Distill analysis steps into precise, falsifiable biological questions with single defensible answers — the core unit of our eval suite.
• Debug platform and data issues with precision: turn messy results across diverse sequencing chemistries into crisp hypotheses, sanity checks, and a stepwise debugging plan.
Requirements
• Experience with end-to-end data analysis for one or more of the following sequencing platforms: MissionBio, ParseBio, CSGenetics, BD Rhapsody, Illumina, or 10X Chromium
• Analyzed 3+ datasets from raw data to end insight for either publications or industry experiments with real world consequences
• Working understanding of platform-specific quality control thresholds and intuition for numerical examples of positive or negative results (e.g., 100K cells from a ParseBio run with 80% mitochondrial reads means something is wrong)
• Familiarity with the landscape of computational biology tools for scRNA-seq tasks (e.g., Scanpy/Seurat for core workflows, cell typing frameworks like CellTypist or Azimuth, DE methods like DESeq2 or edgeR)
• Strong understanding of experimental design, hypothesis generation and scientific conclusions from papers using one of the sequencing platforms described
• Ability to distill an analysis step into a precise, falsifiable biological question with a single defensible answer
• Working understanding of concepts in statistical inference: hypothesis testing, confidence intervals and/or estimators
• Working understanding of important algorithms in high dimensional data analysis: e.g. PCA, neighborhood graphs, UMAP, clustering methods (Leiden/Louvain)
Nice-to-haves
• Published research that relied on modern single-cell RNA sequencing techniques.
• Engineered tools or packages in the single-cell biology domain.
• Experience generating training data for AI agents or foundation models.
Benefits
• $130k–$180k/yr (performance-based)
• Equity
• Unlimited PTO (truly)
• Waterfront office in China Basin, San Francisco
• Free lunch and dinner
• 100% premium covered on Blue Shield's platinum health plan ($0 premium, $0 deductible)
• 401(k) plan options
• Work visa sponsorship
• Company-sponsored professional development
Apply tot his job
Apply To this Job
At LatchBio, our AI agents help thousands of scientists analyze and interpret data across the full stack of modern multi-omic technologies — starting with single-cell and spatial, and expanding fast.
We're building the ground truth for AI in single-cell biology. Our benchmark scBench — 394 verifiable problems across six sequencing platforms — shows the best frontier model today still fails nearly half the time. We're hiring bioinformatics engineers to close that gap: scientists who can turn real experimental data into the precise, falsifiable questions that define what it means for an AI agent to actually understand scRNA-seq.
Responsibilities
• Own end-to-end scRNA-seq analyses across multiple projects: raw platform outputs → QC and failure diagnosis → normalization → dimensionality reduction → clustering → cell typing → differential expression → trajectory analysis → defended biological claim.
• Build reproducible workflows and produce clear decision traces: what was filtered, why, what changed the conclusion, what would falsify the claim.
• Distill analysis steps into precise, falsifiable biological questions with single defensible answers — the core unit of our eval suite.
• Debug platform and data issues with precision: turn messy results across diverse sequencing chemistries into crisp hypotheses, sanity checks, and a stepwise debugging plan.
Requirements
• Experience with end-to-end data analysis for one or more of the following sequencing platforms: MissionBio, ParseBio, CSGenetics, BD Rhapsody, Illumina, or 10X Chromium
• Analyzed 3+ datasets from raw data to end insight for either publications or industry experiments with real world consequences
• Working understanding of platform-specific quality control thresholds and intuition for numerical examples of positive or negative results (e.g., 100K cells from a ParseBio run with 80% mitochondrial reads means something is wrong)
• Familiarity with the landscape of computational biology tools for scRNA-seq tasks (e.g., Scanpy/Seurat for core workflows, cell typing frameworks like CellTypist or Azimuth, DE methods like DESeq2 or edgeR)
• Strong understanding of experimental design, hypothesis generation and scientific conclusions from papers using one of the sequencing platforms described
• Ability to distill an analysis step into a precise, falsifiable biological question with a single defensible answer
• Working understanding of concepts in statistical inference: hypothesis testing, confidence intervals and/or estimators
• Working understanding of important algorithms in high dimensional data analysis: e.g. PCA, neighborhood graphs, UMAP, clustering methods (Leiden/Louvain)
Nice-to-haves
• Published research that relied on modern single-cell RNA sequencing techniques.
• Engineered tools or packages in the single-cell biology domain.
• Experience generating training data for AI agents or foundation models.
Benefits
• $130k–$180k/yr (performance-based)
• Equity
• Unlimited PTO (truly)
• Waterfront office in China Basin, San Francisco
• Free lunch and dinner
• 100% premium covered on Blue Shield's platinum health plan ($0 premium, $0 deductible)
• 401(k) plan options
• Work visa sponsorship
• Company-sponsored professional development
Apply tot his job
Apply To this Job