FSP Associate Manager, Safety Data and Systems - Pharmacovigilance

Summary

Thermo Fisher Scientific is seeking an FSP Associate Manager, Safety Data and Systems to support pharmacovigilance operations. The role designs and validates AI/ML and NLP solutions, manages safety data systems such as Oracle Argus, ensures GxP and regulatory compliance, and collaborates with clinical, regulatory, quality, and vendor teams.

Responsibilities

• Design, develop and validate AI/ML and NLP components that support safety operations - including MedDRA/WHODrug auto-coding, case triage, duplicate detection and narrative summarization - with clear human-in-the-loop checkpoints
• Contribute to model lifecycle management for safety-relevant AI/ML: versioning, monitoring, drift detection, retraining and documentation aligned with GxP / GAMP 5 and internal model governance
• Support the qualification of AI/ML solutions against evolving regulatory expectations (EMA reflection paper on AI, FDA AI/ML guidance, EU AI Act obligations for high-risk systems) in partnership with Quality, DT/BIS and GPS Signal Management
• Serve as the key technical resource for the configuration, maintenance, and administration of the Oracle Argus Safety system
• Support day-to-day operation and troubleshooting of safety systems
• Assist in system validation, testing, and deployment of safety systems updates
• Generate, validate, and customize safety reports and analytics
• Collaborate closely with the pharmacovigilance, clinical, and regulatory teams to ensure safety data management aligns with global regulatory standards (FDA, EMA, PMDA, ICH)
• Participate in change management processes to enhance safety system integrations
• Contribute to audit readiness activities, including system inspections, validation reports, and compliance documentation
• Collaborates with internal systems team, BIS/ DT and Safety vendor on issues related to Safety data
• Performs the generation and quality control of aggregate reports and line listings
• Initiates and contributes to the development of procedural documents including but not limited to Safety Management Plans, SOPs, work instructions, job aides, forms, or templates
• Collaborates and co-creates with applicable client functions (e.g. Medical Information, Data Management, Business Information Systems, Quantitative Science) in regards to pharmacovigilance technical aspects, setup and operation
• Keeps up-to-date on applicable regulatory and PV tech guidelines and shares within GPS and client as applicable
• Participates in training related to safety data management
• Proactively reviews processes and tools and provides suggestions for improvement and better efficiencies
• Complete additional task and projects as assigned by line manager or delegate

Skills

• At least Bachelors' degree (or country equivalent) in computer science, data science, computational linguistics, applied statistics/biostatistics, life sciences / Information technology or other relevant field required
• Python, ML/NLP frameworks, model deployment/monitoring, MLOps tooling, ideally exposure to LLMs on unstructured clinical/safety text
• Working understanding of safety database data models (Argus/ArisG) and E2B(R3) structure
• Relevant experience in IT / Safety / Clinical Research / Pharmacovigilance overall with at least 3 years of proven experience with safety database systems (e.g. ARGUS or ArisG) including workflow management
• Equivalent and adequate combination of education and experience or proven practical expertise in all of the required skills
• Proficiency in Python for ML development, including scikit-learn, pandas, NumPy; experience with at least one deep learning framework (PyTorch or TensorFlow)
• Natural language processing for extraction of adverse events, drugs, and outcomes from unstructured text - case narratives, medical literature, call transcripts, and spontaneous reports
• Named Entity Recognition (NER), relation extraction, and text classification
• Experience with transformer-based / large language models (BERT-family, clinical/biomedical models such as BioBERT or PubMedBERT, and modern LLMs) for narrative generation, summarization, and information extraction
• Supervised and unsupervised methods for classification, clustering, and anomaly detection
• Feature engineering and model evaluation (precision/recall trade-offs, ROC/AUC, calibration) with an understanding of why recall and sensitivity are weighted heavily in a safety context
• Model lifecycle management: versioning, monitoring, drift detection, retraining pipelines using standard MLOps tooling (e.g. MLflow, Azure ML, Databricks) in line with client DT/BIS standards
• Model explainability / interpretability (SHAP, LIME) - essential where decisions must be defensible to health authorities
• Proficiency in electronic systems commonly used for Safety / PV, like for data visualization and analysis, dashboards
• Solid understanding of the quality management processes, metrics and KPIs
• Good knowledge of relevant pharmacovigilance regulatory requirements and guidance documents (including Europe, US, Japan)
• Proficient in the Microsoft 365 stack (Excel, Word, PowerPoint, Teams, SharePoint, OneDrive) and in modern collaboration and documentation tooling
• Advanced Excel required; working proficiency in SQL required for querying safety and operational datasets
• Ability to communicate effectively and collaborate successfully across functions and with vendors
• Fluent communication in written and spoken English required
• Ability to work independently with minimal oversight and prioritize effectively
• Ability to complete multiple complex deliverables within tight timelines
• Ability to function effectively in a team environment
• Familiarity with, or ability to rapidly acquire, GVP/21 CFR 314 concepts preferred
• MedDRA and WHODrug auto-coding using ML/NLP; prompt engineering and retrieval-augmented generation (RAG) a plus
• Understanding of GxP / GAMP 5 validation as applied to AI/ML systems, model governance, and emerging regulatory expectations (EMA reflection paper on AI, FDA guidance) — rare and worth flagging as preferred
• Proficiency in Safety Database systems (e.g. Argus) and knowledge of other technical systems applicable to Safety /Pharmacovigilance (e.g. E2B gateway, safety signal detection tools and systems) is a plus

Qualifications

Must Haves

• At least Bachelors' degree (or country equivalent) in computer science, data science, computational linguistics, applied statistics/biostatistics, life sciences / Information technology or other relevant field required
• Python, ML/NLP frameworks, model deployment/monitoring, MLOps tooling, ideally exposure to LLMs on unstructured clinical/safety text
• working understanding of safety database data models (Argus/ArisG) and E2B(R3) structure
• Relevant experience in IT / Safety / Clinical Research / Pharmacovigilance overall with at least 3 years of proven experience with safety database systems (e.g. ARGUS or ArisG) including workflow management
• Equivalent and adequate combination of education and experience or proven practical expertise in all of the required skills
• Proficiency in Python for ML development, including scikit-learn, pandas, NumPy; experience with at least one deep learning framework (PyTorch or TensorFlow)
• Natural language processing for extraction of adverse events, drugs, and outcomes from unstructured text - case narratives, medical literature, call transcripts, and spontaneous reports
• Named Entity Recognition (NER), relation extraction, and text classification
• Experience with transformer-based / large language models (BERT-family, clinical/biomedical models such as BioBERT or PubMedBERT, and modern LLMs) for narrative generation, summarization, and information extraction
• Supervised and unsupervised methods for classification, clustering, and anomaly detection
• Feature engineering and model evaluation (precision/recall trade-offs, ROC/AUC, calibration) with an understanding of why recall and sensitivity are weighted heavily in a safety context
• Model lifecycle management: versioning, monitoring, drift detection, retraining pipelines using standard MLOps tooling (e.g. MLflow, Azure ML, Databricks) in line with client DT/BIS standards
• Model explainability / interpretability (SHAP, LIME) - essential where decisions must be defensible to health authorities
• Proficiency in electronic systems commonly used for Safety / PV, like for data visualization and analysis, dashboards
• Solid understanding of the quality management processes, metrics and KPIs
• Good knowledge of relevant pharmacovigilance regulatory requirements and guidance documents (including Europe, US, Japan)
• Proficient in the Microsoft 365 stack (Excel, Word, PowerPoint, Teams, SharePoint, OneDrive) and in modern collaboration and documentation tooling
• Advanced Excel required; working proficiency in SQL required for querying safety and operational datasets
• Ability to communicate effectively and collaborate successfully across functions and with vendors
• Fluent communication in written and spoken English required
• Ability to work independently with minimal oversight and prioritize effectively
• Ability to complete multiple complex deliverables within tight timelines
• Ability to function effectively in a team environment

Nice to Haves

• Familiarity with, or ability to rapidly acquire, GVP/21 CFR 314 concepts preferred
• MedDRA and WHODrug auto-coding using ML/NLP; prompt engineering and retrieval-augmented generation (RAG) a plus
• Understanding of GxP / GAMP 5 validation as applied to AI/ML systems, model governance, and emerging regulatory expectations (EMA reflection paper on AI, FDA guidance) — rare and worth flagging as preferred
• Proficiency in Safety Database systems (e.g. Argus) and knowledge of other technical systems applicable to Safety /Pharmacovigilance (e.g. E2B gateway, safety signal detection tools and systems) is a plus

Benefits

• Fully remote work arrangement in North Carolina