Key Responsibilities
- Rapidly prototype AI solutions, validate hypotheses with users/data, and iterate from proof-of-concept to MVP and production-ready releases with measurable outcomes.
- Develop and maintain LLM-powered/agentic applications with human-in-the-loop review, controlled model updates, monitoring/observability, and governance (evaluation, auditability, access controls).
- Build scalable, reliable, secure applications using Python and/or Java and modern front-end frameworks (e.g., React).
- Design and operate secure AWS document ingestion + vectorization workflows (S3, event-driven processing) for governed retrieval across structured/unstructured Regulatory Affairs content (e.g., Amazon S3 Vectors, OpenSearch).
- Design/implement microservices and APIs on AWS using event-driven/asynchronous architectures.
- Implement CI/CD pipelines and infrastructure-as-code for repeatable, compliant deployments.
- Monitor production model performance; manage drift detection, evaluation, retraining/refresh, and rollback strategies.
- Investigate issues across systems/data; deliver short-term fixes and long-term architectural improvements.
Qualifications & Required/Preferred Skills
- 2+ years related software development experience (Pharma/Life Sciences a plus); Bachelorβs in CS/Software Engineering or related.
- Proficiency in Python and/or Java; UI framework preferably React.
- Solid AI concepts; practical experience with generative/agentic AI tools (Claude/OpenAI/LangChain) (plus).
- AWS services (Lambda, S3, DynamoDB, Bedrock, SageMaker) (plus).
- Experience with document ingestion/vectorization for RAG/search and familiarity with vector storage/retrieval (e.g., S3 Vectors/OpenSearch/pgvector).
- Strong SQL/NoSQL analytics; strong problem-solving; collaboration/communication in a matrix organization.
Benefits (as stated)
- Health coverage; wellbeing programs; 401(k), disability/life/other insurance; Paid Time Off; flexible time off and paid holidays (varies by location/role).
Application instruction
- If the role is a near match, apply anyway.