OCI 1Z0-1110-25 Glossary: Key Terms
March 31, 2026
OCI 1Z0-1110-25 glossary of data prep, notebooks, training, deployment, and lifecycle terms.
Use this glossary to clean up high-confusion OCI data-science terms before you go back into mixed sets. On this exam, terminology mistakes usually hide a lifecycle-stage or object-choice mistake.
High-value terms
| Term |
What it means here |
Why it matters on the exam |
| Artifact |
a stored output from a workflow such as a dataset snapshot, model, or notebook result |
artifacts anchor repeatability and traceability |
| Dataset |
the collection of records used for exploration, training, validation, or inference input |
confusion here often causes leakage or split mistakes |
| Deployment |
the stage where a trained model is exposed to serve predictions |
deployment is a lifecycle stage, not just a stored object |
| Experiment |
a tracked run or workflow iteration used to compare behavior |
strong answers separate exploration from repeatable delivery |
| Feature engineering |
preparing and shaping raw data into useful model inputs |
this belongs before reliable training and evaluation |
| Inference |
applying a trained model to new data to generate predictions |
training and inference are not interchangeable |
| Job run |
a scheduled or triggered execution of a data or ML task |
jobs usually signal repeatability, not ad hoc work |
| Model artifact |
the saved trained model output that can later be evaluated or deployed |
artifact existence alone does not prove serving readiness |
| Notebook session |
an interactive workspace for exploration and iterative work |
notebooks are useful, but they are not the whole platform strategy |
| Observability |
the ability to monitor jobs, deployments, and supporting services |
operations questions often hinge on this rather than on model theory |
Common confusion pairs
| Pair |
Clean separation |
| Training vs inference |
training builds or updates the model, inference uses the model on new inputs |
| Notebook work vs production workflow |
notebook work is interactive and exploratory, production workflow is more repeatable and operationalized |
| Artifact vs dataset |
an artifact is a saved output of work, a dataset is the underlying data used or produced |
| Model quality vs platform health |
model quality is about prediction usefulness, platform health is about whether the workflow runs reliably |
| Batch processing vs deployed endpoint |
batch processing handles grouped or scheduled work, a deployed endpoint responds to new requests |
| Validation split vs metric |
the split defines what evidence you trust, the metric defines how you score it |
| Drift vs outage |
drift is quality change over time, outage is service unavailability or failure |
Fast recall anchors
| If you see… |
Think… |
| interactive exploration |
notebook session |
| repeatable scheduled execution |
job run |
| saved trained output |
model artifact |
| live prediction path |
deployment |
If three terms blur together
| Terms |
Short reset |
| project, notebook, job |
project organizes work, notebook is interactive work, job is repeatable execution |
| model artifact, deployment, inference |
artifact is the stored model output, deployment exposes it, inference is the act of using it |
| evaluation, validation, metric |
validation is how you test, metric is how you score, evaluation is the broader judgment |
| model quality, observability, rollback |
quality measures usefulness, observability tells you what is happening, rollback is the recovery move |
Route misses well
| If you missed because… |
Go next |
| you mixed up lifecycle objects |
FAQ |
| you need fast tie-breaks and stage cues |
Cheat Sheet |
| you need a paced rebuild of the weak lane |
Study Plan |
| you need the official Oracle or OCI source |
Resources |
Revised on Monday, June 15, 2026