Website:
flahyhealth.com
Job details:
Flahy is a clinical intelligence layer that processes biological and clinical signals for personalized prevention, early detection, and treatment selection. More information is available at flahyhealth.com
The Role
You'll be responsible for production of bioinformatics pipelines end to end: the compute that takes raw sequencing and genotyping data through to interpreted, reportable output. The workloads are large, embarrassingly parallel in places and stubbornly serial in others, and they have to be reproducible — the same input has to give the same answer six months and three infrastructure changes later. The candidate should be genuinely strong on both halves: the engineering — cloud, containers, parallel execution, cost and performance — and the genomics, deeply enough to know when a pipeline is producing numbers that can't be true. This is a hands-on IC role. You will write the pipelines.
Responsibilities
- Design, build, and operate genomic analysis pipelines from raw reads through alignment, variant calling, QC, and annotation.
- Architect these workloads for parallel execution — scatter-gather at the workflow level, correct use of multiprocessing and threading within tools, sensible sharding of data.
- Handle the cloud infrastructure the pipelines run on: compute provisioning, object storage layout, job orchestration, spot/preemptible strategy, and the cost per sample that results from all of it.
- Containerize every step and keep environments pinned and reproducible.
- Profile and fix the real bottlenecks — I/O, memory, and scheduler overhead are usually the problem before CPU is.
- Build QC that catches bad samples and bad runs before anything downstream consumes them: coverage and uniformity, contamination, concordance, batch effects.
- Maintain annotation sources and interpretation logic with clear provenance for every call that reaches an output.
- Write documentation and validation records that hold up to external review.
Candidate Profile
- 3–6 years building production bioinformatics or scientific computing pipelines — not only research or one-off analysis.
- Strong cloud experience: compute, storage, IAM, managed batch services, and an instinct for what things cost.
- Containerization as a default — Docker, and ideally Singularity/Apptainer in shared-compute settings.
- Real parallel and distributed computing experience: HPC schedulers (Slurm, SGE) and/or cloud batch, plus an understanding of where parallelism actually helps and where it just adds coordination cost.
- Workflow orchestration in Nextflow, WDL, or Snakemake, with genuine reproducibility discipline.
- Strong Python; fluent in the shell; competent with SQL. Comfortable reading C/C++ or Rust when a tool misbehaves.
- Git-based collaborative development, CI, and testing habits that extend to data-producing code.
Genomics
- Genomics fundamentals: variant representation and normalization, reference builds and liftover, coverage and call quality, inheritance patterns, and how population structure affects frequency-based filtering.
- Command of the standard toolchain: BWA/minimap2, GATK and/or DeepVariant, samtools/bcftools, VEP or equivalent annotators — including where each one's defaults are wrong for a given assay.
A Plus
- Molecular biology: you understand what the assay physically does, not just the file format it produces: library prep, amplification, capture chemistry, and the artifacts each step introduces.
- Somatic analysis — tumor-normal or tumor-only, CNV, TMB/MSI.
- Genotyping array processing and imputation at scale.
- GPU-accelerated genomics (Parabricks or similar).
- Kubernetes, Terraform, or equivalent infrastructure-as-code.
- Pharmacogenomics (CPIC, PharmCAT or equivalent).
- Prior work inside an accredited or regulated lab environment (NABL, ISO 15189, CAP/CLIA) — you know what validation and change control actually cost.
- Clinical data standards: HL7/FHIR, HGVS, HPO.
Click on Apply to know more.
