Microsoft's Commerce Risk Applied Science team is seeking to automate manual workflows within their fraud decisioning systems to drive technical innovation, reduce manual intervention, and improve risk assessment accuracy.
Requirements
- 1+ year(s) Experience with feature engineering, model evaluation, and data pipeline design.
- 1+ year(s)Experience with model evaluation platforms.
- Hands-on work with machine learning frameworks and techniques.
- Applied research or development in areas like LLM fine-tuning, evaluation, or RAG implementations.
- Proven track record of delivering scalable and ethical AI solutions.
- Familiarity with manual review workflows and fraud detection systems.
- Prior exposure to vendor data integration and cost-efficiency initiatives.
Responsibilities
- Familiarity of Design and integration of third-party APIs
- Develop and maintain low-latency data ingestion pipelines to feed ML models
- Transform raw API responses into usable model features via middle-layer DAs.
- Collaborate with evaluation teams to invoke and assess models
- Develop, deploy, and maintain machine learning models at scale
- Conduct comparative experiments across APIs to identify valuable data fields and estimate event volumes.
- Analyze manual review cases to define automation scope and edge-case handling strategies.
Other
- Act as the primary point of contact for data science inquiries related to the manual workflow automation
- Coordinate upstream/downstream data needs and manage deliverables via Azure DevOps.
- Align technical execution with business goals, including reducing vendor reliance and savings in manual effort.
- Contribute to achieving maximum automation of manual workflows in Azure, Office, and Consumer business
- Ability to manage ambiguity and drive clarity in complex, cross-functional environments.
