Staff+ Software Engineer, Account Abuse (Machine Learning)

Written by the T&S Network from Anthropic's posting.

Builds machine learning systems that detect and stop account-level abuse and fraud at scale on Anthropic's platform. The Account Abuse team allocates computing capacity fairly, reduces what bad actors can use, and works to keep them from coming back.

• Fraud and scams

What you'd do

• Build and operate a feature platform serving both model training and real-time scoring.
• Create models that flag account abuse and fraud, then run them in batch and live.
• Put backtesting, shadow runs, and phased releases first on the path to production.
• Monitor models for training/serving skew, drift, and adversarial adaptation.
• Build tooling that automates parts of the model lifecycle, using Claude to speed up feature work, training, and evaluation.
• Partner with product and platform teams to integrate model decisions with minimal latency and architecture impact.

What it asks for

• Proficiency in Python and SQL.
• Hands-on history of training ML models and shipping them to production.
• Experience building batch data pipelines with an engine such as Spark or Beam and a scheduler such as Airflow.
• Solid grasp of keeping training data point-in-time correct and avoiding training/serving skew.
• Strong communication skills for explaining tradeoffs to non-technical stakeholders.
• Experience with a feature platform such as Chronon, Feast, or Tecton (preferred).
• Experience with stream processing such as Flink, Beam/Dataflow, or Kafka Streams (preferred).

Our best summary of the posting, and it can be wrong. The employer's posting is what counts: read it in full on Anthropicnorth_east before you apply. If we got something wrong, report this listing.