A 4Shadow Analytix Selected Submission Prepared for Black Diamond Networks August 28, 2026
John Holstein, RPh Regulated science career trajectory
Senior Scientist · Pharmaceutical AI and Safety Knowledge Systems
A medicine is never just a molecule. Pluvicto reveals the full system behind a regulated scientific decision. Follow the drug, the evidence, or human authority through one connected pharmaceutical lifecycle.
See the assignment fit ↓ 40+ years regulated science23 years nuclear pharmacy300+ hospitals in diversion MLCurrent agentic AI delivery
Why this career fits the assignment The JD asks for intersections. Select an intersection to compare the assignment with the career evidence.
01 Agentic AI NewsRx + Amgen 02 Safety intelligence Cardinal + Invistics 03 Knowledge architecture J&J + Cigna 04 Regulated evidence Amgen + Nuclear pharmacy 05 Model evaluation NewsRx + Optinosis 06 Hands-on delivery Current work Agentic AI NewsRx + Amgen Production RAG, multi-agent workflows, structured evaluation and human sign-off.
Job requirement Design and apply agentic AI to pharmaceutical safety work without removing scientific accountability.
Career evidence NewsRx production evaluation and Amgen’s controlled regulatory GenAI join retrieval, judges, feedback and qualified review.
One drug. Three ways through the system.
The complexity is the point. Choose a guided path, then select any stage to inspect its evidence, authority, AI boundary and career proof.
Follow the drug Target to monitored patient dose Follow the evidence Source record to lifecycle decision Follow human authority Where automation stops Play path ▶
PLUVICTO lutetium Lu 177 vipivotide tetraxetan
71 Lu Lutetium
01 Disease and target PSMA-positive prostate cancer 02 Patient selection PSMA PET makes eligibility visible 03 Therapeutic construction Ligand + chelator + lutetium-177 04 Clinical evidence VISION and an expanding indication 05 Regulatory decision Evidence becomes an authorized use 06 Manufacture and release The clock starts before shipment 07 Decay-governed logistics Time, activity and destination are inseparable 08 Right patient, right dose The final handoff is clinical 09 Safety surveillance Every dose creates new evidence 10 Lifecycle learning The graph changes with the medicine
01 PSMA-positive prostate cancer
Disease and target The chain begins with disease state, prior therapy and tumor expression. The therapeutic hypothesis is explicit before data engineering begins.
Human authority Oncology and translational-science leaders
Governed evidence Target profile, protocol rationale and eligibility criteria
AI contribution Connects biological evidence and finds contradictions. It does not establish the indication.
Career evidence DARPA scientific modeling began with physical mechanism and measurable evidence. TPP FOLH1 mCRPC
The knowledge architecture Every acronym has a job. Follow one adverse event as clinical language becomes structured evidence, a safety signal and an accountable lifecycle decision.
01 AE Adverse event → 02 MedDRA Medical Dictionary for Regulatory Activities → 03 ICSR Individual Case Safety Report → 04 PV Pharmacovigilance signal → 05 RMP Risk management and labeling Play chain ▶
01 Physical reality Patient · product · dose · facility
02 Observed evidence Clinical event · assay · batch · exposure
03 Semantic meaning MedDRA · ontology · terminology · provenance
04 Computable knowledge ICSR corpus · comparison · retrieval · signal
05 Accountable decision Validate · report · mitigate · revise
AE Adverse event The observed patient experience becomes a record.
Evidence entering Patient, product, dose, timing and clinical observation
Accountable authority Clinician, pharmacist and safety intake
What leaves this step A traceable case ready for medical review
Career evidence Cardinal clinical and nuclear-pharmacy practice; adverse-event, complaint and exposure reporting. ICH M3(R2) → ICH M4 Development evidence becomes the Common Technical Document. M3(R2) governs when nonclinical evidence supports exposure. M4 gives the complete evidence its reviewable structure.
01 First human exposure Core safety pharmacology and toxicology justify entry into humans. 02 Longer and larger trials Duration, reproductive and other studies expand with exposure and population. 03 Marketing application The package supports benefit-risk review and proposed labeling. 04 Postapproval change New indications, formulations and signals can reopen the evidence argument.
Illuminate career evidence All evidence Cardinal / Syncor Amgen J&J NewsRx / Optinosis
Module 1 Administrative and prescribing information Not harmonized Module 2 CTD overviews and summaries Quality overall summary Nonclinical overview Nonclinical summary Clinical overview Clinical summary
Module 3 Quality data Module 4 Nonclinical study reports Module 5 Clinical study reports
05 Career CTD footprint
All evidence The full career spans prescribing and operational controls, CTD synthesis, quality systems, nonclinical safety, and clinical evidence.
Module 1 Career evidence available Module 2 Career evidence available Module 3 Career evidence available Module 4 Career evidence available Module 5 Career evidence available
Select an illuminated CTD module to read the supporting career evidence.
The living hyperknowledge graph Watch the regulated record learn. Each tile adds an information layer. The mesh grows, resolves identities, acquires rules, weighs evidence, assembles reports and anchors decisions to authority.
Records Documents, rows, events, versions and identifiers enter as disconnected observations. Semantics Synonyms resolve, entities converge and controlled terminology gives records shared meaning. Ontology Classes, constraints and permitted relationships reshape the semantic neighborhoods. Evidence Claims gain support, contradiction, confidence, applicability and provenance. Reports Regulatory artifacts assemble from governed evidence while retaining their source paths. Authority Review, release, prescribing and regulatory decisions anchor the graph to accountable people. Embed new package + Play graph ▶
Provenance Semantic identity Ontology constraint Evidence support Contradiction Human authority
Patient 1847 PSMA PET report Treatment order Administration event Pluvicto Batch 24L77 Release assay Shipment record Adverse event Laboratory result Safety case VISION study Records Documents, rows, events, versions and identifiers enter as disconnected observations. 12 nodes · 10 typed relationships · 0 new evidence packages
New source → Parse + chunk → Vector neighbors → Entity resolution → Proposed edges → Human validation → Governed knowledge
The career is the validation set Four eras. One operating method. Select an era to reveal the work, operating context, result and connection forward.
1979–2005 Physics to patient DARPA modeling, nuclear-pharmacy practice, production and national time-critical distribution.
Evidence shown below 2005–2021 Signals to interventions Geospatial measurement, trafficking intervention, remittance and drug-diversion ML.
View this era 2021–2025 Semantics to evidence J&J ontology, Amgen ICH M4 GenAI, payer semantic layers and regulated reporting.
View this era 2025–today Knowledge to AI Biomedical models, grounded retrieval, multi-judge evaluation and SME review.
View this era 1979–2005 Physics to patient Selected-era evidence Representative work DARPA scientific modeling · Tc-99m labeled-white-cell IND · Cardinal/Syncor nuclear pharmacy · PET hub-and-spoke delivery
Operating context Radiation physics, pharmacy practice, FDA, NRC, cGMP, hazardous-material transport and the Five Rights.
Result Converted decaying material and strict clinical obligations into a reliable national operating system.
Connection forward Physical reality, controlled evidence and accountable handoffs became the foundation for every later system.
The through-line Make complex evidence usable without stripping away authority. The operating method is consistent. Establish reality, govern the data, encode meaning, evaluate the model, and preserve the professional decision.
John Holstein, RPh · Pharmaceutical AI and Safety Knowledge Systems A 4Shadow Analytix experience · AI-assisted research and development Return to top ↑