About the role

About Raycaster
We build AI agents for life sciences. Backed by leading life sciences investors and YC, we’re already working with Agilent, Yokogawa Life Sciences and many leading companies to automate their regulated, complex processes to accelerate drug development.

The role (what you’ll do)

• Partner with CMC teams to map current workflows end-to-end (inputs, systems, approvals, artifacts).
• Decompose tech-transfer packages (method revs, redlines/rationales, representative chromatograms, acceptance tests) into steps an AI can execute with citations.
• Capture validation/stability/release processes (accuracy/precision/linearity/LOQ, OOT rules, spec checks, CoAs) and translate them into checklists, schemas, and templates.
• Define data and document requirements (e.g., LIMS/CDS exports, S3/SharePoint/Vault locations), edge cases, and “fail-closed” rules (no source → no claim).
• Review AI-generated drafts for scientific correctness and compliance expectations (21 CFR Part 11).
• Create clear SOPs/playbooks for each automated workflow; iterate with engineers to reach “ready for pilot” quality.

You might be a fit if you have

• 3 years in CMC/Analytical Development or QC, ideally touching Phase 2/3 or tech transfer to CRO/CMO/CDMO.
• Hands-on with HPLC/LC-MS (Ion-Pair, HILIC, AEX), method/SOP authoring, and interpreting system-suitability & chromatograms.
• Built or reviewed validation packets (ICH Q2(R1/R2)), stability trending/OOT, release specs/CoAs, and change control narratives.
• Operated in non-GMP → GMP handoffs; practical understanding of 21 CFR Part 11 expectations (audit trail, e-sig routing, data integrity).
• Familiarity with at least some of: Veeva Vault (or MasterControl), ELN/LIMS (Benchling, LabWare, SampleManager), CDS (Empower/Chromeleon), Jira/ServiceNow.
• Comfort collaborating with engineers; you enjoy turning tacit know-how into templates, schemas, and acceptance criteria.

Nice to have

• ASO, mRNA/LNP, peptides, or oligo experience; Phase 3/registration exposure.
• Awareness of CSV (computerized system validation), ALCOA+, and doc control practices.
• Light data skills (reading CSVs, understanding units/columns) and curiosity about AWS/S3/Athena-style data lakes.

Why this is interesting
You’ll help design the cutting edge of audited AI workflows that save teams hours per artifact while keeping reviewers in control. Your fingerprints will be on how modern CMC gets done.

About the interview

1. Take-home assignment (30 min) to assess existing expert knowledge.
2. Workflow deep-dive (30 min): You walk us through a real late-stage workflow you know cold (e.g., tech transfer pack or validation summary). We map inputs → outputs → approvals and capture edge cases.
3. Offer

About Raycaster

Raycaster builds AI systems for complex professional work and helps leading AI labs train and evaluate models on that work. We began as an agentic AI platform serving dozens of publicly traded companies across life sciences, energy, and oil and gas. Building on that experience, we launched a division focused on domain-specific training data, benchmarks, and evaluation environments.

Our work brings together engineers, researchers, and industry experts to turn real workflows into measurable improvements in AI capability. Joining Raycaster means working on both sides of the problem: building agents that perform useful work and developing the evaluations that reveal where they still fall short.