The design (entities, sources, views, TTLs, services) transfers across platforms; the vocabulary differs.
| Concept | Feast | Tecton | Databricks (Feature Engineering in Unity Catalog) | SageMaker Feature Store | Vertex AI Feature Store | Hopsworks |
|---|---|---|---|---|---|---|
| Entity / key | Entity(join_keys=...) | Entity(join_keys=...) | primary key columns of a feature table | record identifier | entity ID column | primary key |
| Group of features | FeatureView | BatchFeatureView, StreamFeatureView | feature table (Delta) | feature group | feature view over BigQuery | feature group |
| Event time | timestamp_field | timestamp_field | timestamp_keys (time series feature table) | event time feature name | timestamp column | event_time |
| TTL | ttl | ttl / serving_ttl | lookback window in FeatureLookup | online TTL | — (latest value) | TTL on online store |
| Online store | Redis, DynamoDB, SQLite... | built-in (DynamoDB/Redis) | online tables / Lakebase | online store | Bigtable / optimized online serving | RonDB |
| Request-time features | on_demand_feature_view | RealtimeFeatureView | feature functions (UC functions) | — (compute in the app) | — | on-demand transformation functions |
| Model's feature list | FeatureService | FeatureService | FeatureLookup list in a training set | — | feature view selection | feature view |
| Historical retrieval | get_historical_features | get_features_for_events | create_training_set(... timestamp_lookup_key=...) | Athena/Spark query over offline store | BigQuery point-in-time query | feature_view.training_data() |
Databricks example
from databricks.feature_engineering import FeatureEngineeringClient, FeatureLookup
fe = FeatureEngineeringClient()
fe.create_table(name="ml.churn.customer_activity", primary_keys=["customer_id"], timestamp_keys=["event_ts"],
df=activity_df, description="Rolling engagement aggregates")
training_set = fe.create_training_set(
df=spine_df, # customer_id, score_ts, label
feature_lookups=[FeatureLookup(table_name="ml.churn.customer_activity", lookup_key="customer_id",
timestamp_lookup_key="score_ts", lookback_window=timedelta(days=2))],
label="churned")SageMaker example
from sagemaker.feature_store.feature_group import FeatureGroup
fg = FeatureGroup(name="customer-activity", sagemaker_session=session)
fg.load_feature_definitions(data_frame=activity_df) # infers types
fg.create(s3_uri=f"s3://{bucket}/offline", record_identifier_name="customer_id",
event_time_feature_name="event_ts", role_arn=role, enable_online_store=True)Point-in-time training sets are built by querying the offline store (Athena/Spark) with "latest row with
event_ts <= spine time" logic, or with the SDK's dataset builder (as_of, point_in_time_accurate_join).
Choosing a platform
| Situation | Typical choice |
|---|---|
| Open source, bring your own infra | Feast |
| Heavy real-time/streaming features, managed | Tecton |
| Already on Databricks / Unity Catalog | Databricks Feature Engineering |
| AWS-native ML stack | SageMaker Feature Store |
| GCP / BigQuery-centric | Vertex AI Feature Store |
| Full ML platform with feature store | Hopsworks |
APIs change across versions; check the platform's current docs for exact signatures before generating code for anything other than Feast.
anomaly-detection-model-builder
Build and evaluate anomaly detection for tabular, sensor and transaction data with few or no labels: define the anomaly type (point, contextual, collective), choose detectors (robust z-score/MAD and IQR rules, Mahalanobis with robust covariance, k-NN distance, LOF, Isolation Forest, one-class SVM, PCA reconstruction / autoencoders, forecasting residuals for time series), combine them by rank, explain each alert by its deviating features, set thresholds from an alert budget or the score distribution, and evaluate with precision@k, PR-AUC and recall on known incidents, or detector agreement when no labels exist. Includes pure-Python detectors (robust z, Mahalanobis, kNN, Isolation Forest, ensemble) and an evaluation script.
catboost-model-builder
Build CatBoost models that use its strengths correctly: raw categorical columns as cat_features (ordered target statistics, no manual encoding), text_features, Pool construction with NaN-safe categories, ordered boosting and has_time for temporal data, symmetric (oblivious) trees, overfitting detector with eval_set and use_best_model, losses for classification, regression (Quantile, Poisson, Tweedie) and ranking (YetiRank), class weights, monotone constraints, feature importance (PredictionValuesChange, LossFunctionChange) and SHAP values, GPU training and model export. Plans cat/text/numeric columns from a CSV and generates a runnable training script, and checks parameter sets for the conflicts CatBoost rejects (iterations vs n_estimators, Bayesian bootstrap with subsample, multiple class-weight options, depth > 16, use_best_model without eval_set).
categorical-encoding-selector
Choose and implement the right encoding for every categorical column given the model family and the data: one-hot (with infrequent-level grouping and handle_unknown), ordinal with an explicit order, cross-fitted target encoding (scikit-learn TargetEncoder, CatBoost ordered statistics, category_encoders inside CV), frequency/count encoding, feature hashing, native categorical support in LightGBM, HistGradientBoosting, XGBoost and CatBoost, prefix levels for codes (zip3, NAICS-2), and embeddings. Measures cardinality, rare-level share, unseen-level rate on a time-ordered holdout and target signal per column, emits the ColumnTransformer, and quantifies target-encoding leakage (in-sample vs out-of-fold vs holdout AUC).
classification-model-selector
Choose the right classification algorithm for tabular, text or mixed data and prove the choice with a fair bake-off: logistic regression (L2, L1, elastic net), linear SVM, naive Bayes, k-NN, RBF SVM, decision trees, random forest, extra trees, HistGradientBoosting, LightGBM, XGBoost, CatBoost, explainable boosting machines (EBM/GAM), points scorecards and MLPs, matched to data size, width, categorical and high-cardinality share, missingness, class imbalance, text, and constraints such as interpretability, monotonic constraints, calibrated probabilities, latency budgets and tiny deployable models. Ranks candidates with starting hyperparameters and cautions, then generates a bake-off script running every candidate on identical CV folds with per-fold metrics, fit time and prediction latency.
clustering-algorithm-selector
Choose the right clustering algorithm and check that clustering makes sense at all: cluster tendency (Hopkins statistic with a correlated-features caveat, nearest-neighbour isolation), then k-means / MiniBatchKMeans, Gaussian mixtures with BIC, agglomerative (Ward, average, complete, single) with dendrograms, DBSCAN, HDBSCAN, OPTICS, spectral clustering, BIRCH, mean shift, k-prototypes and k-modes for mixed or categorical data, and Gower distance, matched to data size, expected shapes, noise, varying density, known k, soft memberships, hierarchy and streaming. Covers preprocessing (scaling, encoding, dimensionality, distance metrics), validation (internal indices, bootstrap stability, external labels, business profiling) and assigning new points.
collaborative-filtering-implementer
Implement collaborative filtering recommenders that beat popularity: item-based kNN with cosine, Jaccard or asymmetric similarity, shrinkage, neighbourhood size, positive-only or mean-centred rating signals and 'because you liked X' explanations; implicit ALS with confidence weighting, BPR pairwise ranking and the closed-form EASE model in plain NumPy, tuned on a temporal validation split by recall@k and NDCG@k and refit on all data; plus user-kNN, explicit matrix factorisation, the implicit/LightFM libraries, hyperparameter ranges, cold-start fallbacks, scaling with sparse matrices and approximate nearest neighbours. Produces top-k lists in the format of the recommendation evaluator.