Data Operations Manager: Anthropic

This is not a traditional "Data Engineering" role where you write ETL pipelines in Airflow all day. This is a Supply Chain Management role, but the product is Intelligence.

Anthropic needs massive amounts of highly specific, high-quality data to train Claude (e.g., "Write a Python script to fix this bug" or "Analyze this legal contract"). They cannot generate this all internally. They rely on a vast network of external vendors, human annotators, and subject matter experts.

Your job is to be the General Contractor. You take a request from a PhD Researcher ("We need 10,000 examples of complex reasoning"), figure out how to source it, hire/manage the vendors to create it, ensure the quality is perfect, and deliver it back to the model.

Key Responsibilities
The "Translator": Researchers speak in math and theoretical concepts. You must translate "We need better tool-use capabilities" into concrete instructions for human annotators (e.g., "Create scenarios where the AI must use a calculator and a search engine simultaneously").

Vendor Diplomacy: You will likely manage relationships with companies like Scale AI, Labelbox, or specialized expert networks. You are responsible for the budget, the timeline, and the quality.

Quality Assurance (QA): If the data is bad, the model is bad. You must build systems to catch errors. (e.g., "How do we know the annotator actually solved the math problem correctly?")

Process Scaling: Moving from manual spreadsheets to automated workflows as the data volume explodes.

Strategic Analysis: Why the Pay is So High
The Bottleneck: Compute (GPUs) is available. Algorithms are published. High-quality human data is currently the biggest bottleneck in AI progress.

The "Garbage In, Garbage Out" Risk: A single bad dataset can ruin a training run that costs millions of dollars in compute time. You are the insurance policy against that waste.

The Profile Mismatch: They want someone with the operational rigor of a Management Consultant (McKinsey/Bain) but the technical literacy to understand RLHF (Reinforcement Learning from Human Feedback). Finding people who can do both is difficult.

Candidate Profile
Ideal Backgrounds:

Ex-Consulting: (MBB/Tier 2) capable of structured problem solving.

Tech Ops: (Uber/DoorDash/Airbnb) Operations Managers who have managed complex human-in-the-loop systems.

Program Managers: Technical PMs who have worked with data pipelines.

Key Trait: Ambiguity Tolerance. The researchers often don't know exactly what they need until they see it. You must be comfortable pivoting strategies weekly.

Technical Skills: You don't need to be a Machine Learning Engineer, but you need to know SQL and be able to read/understand Python to debug data issues.