Senior Software Engineer

Adobe•San Jose, CA

About The Position

The Adobe Risk Platform (ARP) is Adobe's centralized fraud prevention and risk-decisioning platform, protecting surfaces like Commerce, Stock, and Firefly with real-time decisions that minimize impact on the customer experience. As a Senior Software Engineer focused on Data Science and Platform Engineering, you'll build the ML models that detect and mitigate risk, fraud, and abuse, and also work on building the low-latency platform that serves them at scale across Adobe's most sensitive, highest-volume surfaces.

Requirements

Bachelor's or Master's in Computer Science, Statistics, ML, or related — or equivalent experience.
Deep experience with ML techniques — supervised/unsupervised learning, anomaly detection, ensembles — applied to fraud or risk detection.
5+ years building production ML systems and scalable backend services.
Comfortable owning a model's full lifecycle in production — monitoring, retraining, and tuning as fraud patterns shift.
Fluent in Python and SQL; comfortable with ML frameworks (PyTorch, TensorFlow, scikit-learn) and a production language (Java, Go, or similar).
Experience with big data tools like Spark or Databricks for training and serving at scale.
Strong platform engineering skills: APIs, distributed services, and data pipelines built for scale and low latency.
Curiosity to understand how abuse actually plays out in a workflow, not just how it looks in aggregate metrics — new patterns often show up in the field before they show up in the data.
Works well across fraud, platform, and product teams.

Nice To Haves

Real-time scoring systems
Device fingerprinting
Using LLMs/agentic AI for fraud detection
Experience working on risk mitigation

Responsibilities

Own the ML lifecycle end-to-end — data engineering, offline model development, and real-time inference — building the models that detect and stop fraud and abuse across our surfaces.
Turn behavioral, device, and transaction signals into features that sharpen detection.
Track precision, recall, false positives, and business impact, and experiment with new approaches (including LLMs and agentic AI) to catch more while reviewers do less.
Own the platform that serves those models in production, from APIs and SDKs to feature stores and the integration points that make onboarding new surfaces turnkey. Keep it fast, reliable, and scalable as usage grows.
Partner with fraud investigators, Trust & Safety, and platform teams to turn new fraud patterns into new defenses, fast.