Staff Scientist

BCC-NIH•Bethesda, MD

23d•$110,000 - $140,000

About The Position

Black Canyon Consulting is seeking a Staff Scientist to work with a Principal Investigator at the National Institutes of Health’s National Library of Medicine. The role supports development of high-fidelity artificial intelligence models to decode the functional landscape of human and mouse genomes, leveraging Telomere-to-Telomere (T2T) reference assemblies to advance understanding of gene regulation, particularly in complex and repetitive genomic regions. The position combines computational genomics, machine learning, and scalable software engineering to support large-scale data integration and model development. The Staff Scientist will develop and scale a cloud-native, TREDNet-based framework; optimize models for distributed, multi-GPU training; integrate genomic and epigenomic datasets; and apply AI methods to annotate repetitive regions such as centromeres and telomeres. The role also involves building containerized pipelines, implementing MLOps practices, deploying workflows in cloud environments, and collaborating across computational and life sciences teams. Deliverables include a containerized TREDNet pipeline scalable across multiple GPU nodes in a cloud environment, a functional map of the T2T reference genome that identifies regulatory motifs in previously unresolved regions, and comparative human and mouse cell-line models to identify conserved regulatory mechanisms.

Requirements

PhD in Computer Science, Computational Biology, Bioinformatics, or a related field.
At least 5 years of experience developing and deploying machine learning or deep learning models.
Strong experience with cloud platforms such as AWS, GCP, or Azure.
Proficiency in deep learning frameworks: PyTorch preferred; TensorFlow or HuggingFace acceptable.
Deep understanding of neural network architectures, including CNNs, transformers, and sequence models.
Strong programming skills in Python and experience working in Linux-based environments.
Experience with MLOps practices, including experiment tracking and model versioning.
Experience building and deploying containerized workflows using Docker and/or Singularity.
Experience with distributed training across GPUs or multi-node environments.
Strong knowledge of genomics, gene regulation, and epigenomics.
Experience working with large-scale biological datasets such as ENCODE, Roadmap Epigenomics, and the UCSC Genome Browser.
Familiarity with genomics data formats, including FASTA, VCF, BAM/CRAM, and BED.

Nice To Haves

Experience with Telomere-to-Telomere (T2T) genome assemblies.
Experience analyzing repetitive genomic regions such as centromeres and telomeres.
Background in regulatory, functional, or comparative genomics, such as human versus mouse.
Experience with hyperparameter tuning and large-scale model optimization.
Familiarity with genomic foundation models or sequence-based deep learning approaches.
Experience running ML workloads on GPU-enabled cloud or HPC environments.
Familiarity with workflow orchestration tools such as Nextflow, Snakemake, or Airflow.
Experience transitioning research models into production-grade pipelines.
Familiarity with CI/CD and infrastructure-as-code tools such as Terraform.
Experience working in interdisciplinary teams.

Responsibilities

Lead the design, development, and implementation of AI-driven models for gene regulation analysis.
Architect and scale a TREDNet-based framework for cloud-native execution.
Optimize models for distributed, multi-GPU training environments.
Integrate and analyze large-scale genomic and epigenomic datasets, including ENCODE/modENCODE, NIH Roadmap Epigenomics, and the UCSC Genome Database.
Apply AI methodologies to functionally annotate repetitive genomic regions, including centromeres and telomeres.
Develop and maintain scalable, containerized pipelines using Docker and/or Singularity.
Implement MLOps practices, including experiment tracking, model versioning, and reproducibility.
Deploy and manage workflows in cloud environments such as AWS, GCP, or Azure.
Collaborate with interdisciplinary teams across computational and life sciences domains.
Develop a containerized TREDNet pipeline capable of scaling across multiple GPU nodes in a cloud environment.
Produce a comprehensive functional map of the T2T reference genome, identifying regulatory motifs in previously unresolved regions.
Develop comparative models between human and mouse cell lines to identify conserved regulatory mechanisms.