Scientific Data Platform Architect - Antibody Discovery

Prellis Bio • Berkeley, CA

31d

About The Position

At Prellis we integrate human biology with machine learning. We aim to revolutionized drug discovery by harnessing the power of human immune system with tightly, integrated machine learning to develop next-generation antibody therapeutics with unparalleled speed, precision and safety. We are committed to empowering our pharmaceutical partners with access to the most promising fully human body candidate rapidly identified from the human immune repertoire, enabling them to bring life-changing treatments to patients faster than ever before. Prellis Biologics is a pre-IPO biotech located in Berkeley CA with a team-oriented, inclusive, and family-friendly culture. Our growing pipeline target high unmet patient needs across therapeutics including metabolic, inflammation, and oncology disease. Prellis has raised funding from top investors, including Celesta, Khosla Ventures, SOSV, & Avidity Partners. You'll architect and hands-on build the end-to-end scientific data platform that powers antibody discovery and characterization. This includes a well-structured PostgreSQL backbone on AWS, reliable ETL from lab systems (Benchling, PipeBio, instruments), and a scientist-friendly app (Shiny or Python) with built-in analytics and visualizations. You'll design for FAIR data (Findable, Accessible, Interoperable, Reusable) and publish AI/ML-ready datasets with clear lineage and versioning.

Requirements

• Bachelors degree is Computer Science or similar field
• 7+ years building data platforms or complex data products; expert SQL/PostgreSQL (schema design, optimization, migrations).
• Strong Python or R for data engineering and app development (Pandas/SQLAlchemy or Shiny/Plotly/Streamlit).
• Proven ETL experience from files/APIs and pragmatic scheduling (cron/Airflow/Prefect-keep it simple).
• Practical AWS with Postgres on RDS/Aurora, S3 for storage, basic IAM/VPC, and CloudWatch for monitoring.
• Hands-on analytics & visualization for scientific datasets.
• Working knowledge of FAIR principles and shaping AI/ML-ready datasets (features, labels, versioned exports).

Nice To Haves

• Benchling developer experience (entities, webhooks) and familiarity with PipeBio outputs.
• Exposure to lab data types (FCS, BLI/SPR, ELISA, NGS summaries, PDB) and data integrity concepts (ALCOA+, 21 CFR Part 11 basics).
• Light containerization (Docker) and deploying a small app on EC2/ECS.
• Experience round-tripping model outputs to a database/UI; comfort with Jupyter/scikit-learn/PyTorch.

Responsibilities

• Platform architecture & data modeling (Postgres on AWS)
• Own the canonical schemas (with selective JSONB), indexing/partitioning, materialized views, and stable entity IDs (samples, sequences, assays, runs).
• Operate RDS/Aurora PostgreSQL, S3 for raw artifacts, and right-sized IAM/VPC access; set guardrails for backups, recovery, and monitoring (CloudWatch).
• FAIR by design & governance
• Make data Findable (catalog/registry tables, searchable metadata), Accessible (role-based access, documented APIs/exports), Interoperable (controlled vocabularies, standard formats such as CSV/Parquet, FASTA/VDJ, FCS/SPR), and Reusable (required metadata, units/QC flags, versioned tables).
• Define and enforce data contracts, provenance, and lightweight review checkpoints.
• Ingestion & transformation (ETL/ELT)
• Build parsers/pipelines for instrument exports (CSV/TSV, FCS, ELISA/SPR/BLI), PipeBio repertoire/QC outputs, and Benchling entities via API/webhooks.
• Add validation, unit normalization, schema migrations, and automated checks.
• Analytics & visualization (data + display layers)
• Create curated analytic views (assay roll-ups, QC dashboards, lineage), and implement interactive visuals (dose-response fits, sensograms, flow summaries, repertoire plots) with Plotly/Dash, Shiny, Spotfire, Streamlit, or similar.
• Deliver drill-downs, comparisons across runs/targets, and clean CSV/Excel exports.
• Application layer
• Build and maintain a small Shiny (R/Python) or Python app (FastAPI + Dash/Plotly/Streamlit) that is role-aware, searchable, and easy for scientists to use; deploy simply (EC2/ECS/Docker).
• AI/ML interface
• Publish feature-ready Parquet/Arrow datasets (sequence features, developability metrics, assay labels like KD/EC50, clonotypes) with dataset versioning, timestamps, and lineage.
• Provide reproducible extracts/snapshots for training, and ingest model predictions/scores back into Postgres and the UI.
• Technical leadership
• Set patterns and code standards, mentor contributors, review designs, and coordinate with Biology, Analytics, and QA/Compliance.
• Keep cost/performance sane; evolve the roadmap as assays and throughput grow.
• Immediate Projects
• A clear Postgres schema with stable IDs, required metadata, and provenance supporting FAIR discovery.
• Automated ETL for Benchling + PipeBio + instruments, with validation and unit normalization.
• A usable app delivering interactive analytics & visualizations scientists rely on daily.
• ML-ready datasets with documented contracts; backups, monitoring, and a published data dictionary/metadata guide

Benefits

• A competitive employee benefits package, including group medical, dental and vision coverage, life and disability insurance, flexible spending accounts an a 401(k) plan
• Stock-based long term incentives
• Bonus plan
• Holiday package including a 1+ week winter shutdown
• Flexible work models, including remote and hybrid working arrangements, where possible

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level
• Industry: Chemical Manufacturing
• Number of Employees: 11-50 employees