Job Description

Job Description

Johnston, IA – candidate living within 50-mile radius of location required onsite T/W/TH each week.

Project Scope and Brief Description:

Work at the intersection of plant cell biology and applied AI to build, productionize, and maintain computer vision pipelines that accelerate Doubled Haploid (DH) breeding in Biotechnology. The contractor will contribute to end‑to‑end imaging and analytics—from microscopy microspore detection to macroscopic structure assessment and plantlet characterization—supporting decisions that reduce cycle time and cost in DH programs. Solutions will be developed primarily in Python, integrated with our repositories and workflow tooling, and aligned with Biotech strategy initiatives.

Responsibilities:

• Design & deliver deep learning-based CV models for microscopy and macroscopic assays (detection, segmentation, classification) with measurable accuracy, robustness, and throughput.
• Build production‑ready pipelines in Python (data ingest, preprocessing, augmentation, inference, batch processing), integrated with GitLab repos and experiment tracking; ensure reproducibility and documentation.
• Implement hyperspectral analysis workflows (band selection, normalization, feature extraction, model training).
• Harmonize imaging acquisition with analysis by collaborating with biology teams to standardize microscopy/RGB/hyperspectral capture and file formats (e.g., FIJI/ImageJ for z‑stacks; autoscale practices).
• Quantify model performance (precision/recall, F1, ROC/AUC, calibration) and write clear reports/posters for DH sessions; support fact‑checking in presentations.
• Operationalize at scale: batch processing of tens of thousands of structures/images; optimize inference (e.g., torch.compile, mixed precision) and monitor resource usage.
• Partner with DH stakeholders (biotech & breeding, Genome Technology Discovery, Data Science) to align deliverables with deployment milestones.
• Maintain IP & data stewardship practices consistent with internal strategy; avoid disclosure of confidential protocols while enabling model re‑use.

Skills / Experience:

Must‑Have

• 4–6 years hands‑on in computer vision with Python (PyTorch/TensorFlow), including detection/segmentation/classification for scientific or industrial imaging.
• Proven ability to productionize models: Git/GitLab, code reviews, CICD basics, experiment tracking (MLFlow or equivalent), reproducible data/experiments, and clear documentation.
• Experience with microscopy image processing, multi‑page TIFFs, z‑stacks, autoscale/normalization, and image quality challenges.
• Familiarity with hyperspectral or multispectral imaging pipelines (preprocessing, dimensionality reduction, modeling) applied to plant or biological materials.
• Track record of measurable model performance reporting and communicating results via posters/presentations for technical audiences.

Nice‑to‑Have

• Vision Transformers (ViT) and modern YOLO workflows for microscopy/macroscopic tasks; comfort with infer tooling.
• Experience optimizing inference (e.g., torch.compile, mixed precision) and scaling batch workflows.
• Domain familiarity with Biotech breeding workflows.
• Collaboration with discovery and strategy teams; ability to work across biology, engineering, and data science groups.

Soft Skills

• Strong stakeholder communication and the ability to translate biology & process constraints into CV requirements; comfortable triaging and prioritizing rapidly in active programs.
• Ownership mindset around documentation, reproducibility, and IP‑aware sharing.
• Curious and learning mindset
• Technical leadership experience.

Job Tags

Similar Jobs