DataRobot Certified Professional study plan

Map the five domains against the exam weighting and rank your weakest sub-skills before studying anything

Define agentic AI in DataRobot terms: agents, tools, tasks, and orchestration loops, and write a one-page glossary in your own words

Identify which agentic components are configured declaratively versus coded, and note where guardrails attach

Build a mental model of how an agent selects and invokes a tool, including failure and retry paths

Trace an end-to-end agent workflow from intake to final answer, naming every handoff point

Compare single-agent and multi-agent patterns and state when each is appropriate

List the failure modes of tool-calling agents and the mitigation for each

Draft five practice scenarios where you choose the correct orchestration pattern and justify it in two sentences

Distinguish generative model deployment, prompt-based workflows, and retrieval-augmented patterns as DataRobot presents them

Explain how vector databases and embedding models fit into a retrieval pipeline and where chunking decisions matter

Describe evaluation approaches for generative outputs, including when human review is required

Identify the guardrail and content-safety controls available for generative endpoints

Explain prompt template management, versioning, and rollback for production generative features

Describe the predictive modeling lifecycle from problem framing through deployment and monitoring

Compare feature engineering, feature discovery, and automated feature generation in the platform

State how model selection and leaderboard ranking work and what the selection metric trade-offs are

Walk through training, tuning, and blueprint selection for a tabular prediction problem

Explain how to interpret model explanations such as feature impact and prediction explanations

Describe retraining triggers and how to decide between retrain, replace, and retire

Identify the failure modes of predictive models, including drift, leakage, and target imbalance

Enumerate the governance artifacts a deployed model must carry, including documentation, ownership, and approval records

Explain how model risk tiers change review requirements and approval paths

Describe how to document a model's intended use, limitations, and known failure modes

Map the governance controls that apply differently to generative versus predictive models

Define the service health, data drift, and model accuracy signals that observability must track

Explain how alerts are configured, routed, and triaged, and what distinguishes a page from a ticket

Describe how to investigate a production degradation from symptom to root cause

Connect observability findings back to governance records and retraining decisions

Revisit every flagged open question from weeks 1 through 7 and resolve or explicitly park each one

Synthesize cross-domain connections, especially governance and observability touching all three model types

Complete timed practice on your two weakest domains and review every miss for the underlying concept

Rehearse explaining each domain's core concepts aloud in under three minutes