Eric’s Field Notes MOC
The recordings Eric made himself between sessions: hallway conversations, a vendor consultation, and the debriefs that turned conference material into open questions and career threads.
Transcripts
- Audit war rooms and mock audits - how FDA inspections actually run, the war-room loop, and Eric’s open question about whether his own team runs mock audits
- Novel acceptance criteria via LLM - Abhishek Bachchan on grounding, generalizing, and originality-checking an acceptance-criteria statement
- Vendor consultation - evaluating an AI/LLM GxP validation platform; Eric’s open question about centralizing KPIs and evals
- Compliance integrity and career pathways - enforcement war stories, Eric’s institutional-trust framing, and his arc from microbiology to clinical data science
Evergreen notes
- Regulatory inspections verify execution against SOPs, not the existence of documents
- Never guess in an inspection; say you will double check
- GxP documentation builds institutional trust on behalf of patients
- Centralize evaluation when every AI pairing carries its own KPIs
- Grounding an LLM in the regulatory corpus yields defensible, novel acceptance criteria
Open threads out of the conference
- The mock-audit question: check with the Clindav digital lead about the mock-audit schedule (from the war-room conversation).
- Centralized evaluation: every LLM paired with an application carries distinct KPIs and scattered evals; the vendor-platform pattern is one shape of the answer.
- Follow-up beyond this repo: Eric’s 2026-09-24 mentorship one-on-one opened with this conference’s debrief (right-sizing rigor to risk, roughly 70 checklist items for simple domains and 150 for hard ones); that note lives in his private vault.