Marketplace Engineering Recsys Feature Engineering Best Practices
Comprehensive first-principles guide for deriving usable recommender features from the raw assets of a two-sided trust marketplace — listing photos, owner-supplied listing metadata, and sitter wizard responses — for item-to-item, user-to-item, and user-to-user solutions. Contains 44 rules across 8 categories ordered by cascade impact on the feature-engineering lifecycle, plus one playbook that composes the rules into an end-to-end feature discovery workflow.
This skill is the upstream precursor to marketplace-personalisation (AWS Personalize) and marketplace-search-recsys-planning (OpenSearch retrieval). Those skills treat features as inputs they already have; this skill is about deciding what features to build from the raw assets, which decisions they serve, and how to prove each one is worth its maintenance cost.
When to Apply
Reference this skill when:
- Planning what to extract from listing photos, descriptions, or amenity lists to power i2i similarity or u2i ranking
- Designing or revising the sitter onboarding wizard with recsys features as the primary output
- Deciding whether to build a vision embedding pipeline, a text encoder, or neither — and in what order
- Composing existing base features into item-to-item, user-to-item, or user-to-user scoring
- Auditing an existing feature store for coverage, drift, PII, duplication, or orphan features
- Choosing a ship/kill criterion for a new recsys feature and designing the ablation A/B test
- Answering the question: "we want to improve the similar-homes shelf — what feature should we build?"
Setup
This skill has no user-specific configuration — it is self-contained. References are live URLs to engineering blogs from Airbnb, Pinterest, DoorDash, Uber, Netflix, and Google, to open-source libraries (Feast, Sentence-Transformers, Hugging Face CLIP, H3), to foundational academic papers (Airbnb KDD 2018, Pinterest ItemSage, YouTube Semantic IDs, PinSage), and to Google's Rules of Machine Learning.
Rule Categories
Categories are ordered by cascade impact on the feature-engineering lifecycle: auditing mistakes build features on data that does not exist, first-principles mistakes produce features that do not map to real decisions, extraction mistakes poison everything downstream, and so on. Fix earlier-stage problems before later-stage problems.
| # |
Category |
Prefix |
Impact |
| 1 |
Asset Audit and Inventory |
audit- |
CRITICAL |
| 2 |
First-Principles Feature Decomposition |
firstp- |
CRITICAL |
| 3 |
Image Feature Extraction |
vision- |
HIGH |
| 4 |
Listing Text and Metadata Extraction |
listing- |
HIGH |
| 5 |
Sitter Wizard and Profile Extraction |
wizard- |
HIGH |
| 6 |
Derived Similarity and Affinity |
derive- |
MEDIUM-HIGH |
| 7 |
Feature Quality and Governance |
quality- |
MEDIUM-HIGH |
| 8 |
Incremental Rollout and Value Proof |
prove- |
MEDIUM |
Quick Reference
1. Asset Audit and Inventory (CRITICAL)
audit-measure-coverage-before-modelling — reject fields below 80% coverage from the feature plan
audit-sample-every-asset-type-end-to-end — pull 100 real instances through the real fetch path before planning
audit-verify-rights-and-privacy-before-extraction — ToS, GDPR, consent, face blur before encoding
audit-quantify-freshness-per-asset — age distribution + expiry + refresh bucket
audit-separate-raw-assets-from-derived-features — raw immutable in object store, derived versioned in feature store
2. First-Principles Feature Decomposition (CRITICAL)
firstp-start-from-the-decision-not-the-algorithm — decision first, sub-judgments second, tools last
firstp-ask-what-signal-a-human-uses — interview 8-12 owners and sitters; features trace back to quotes
firstp-tie-every-feature-to-a-specific-solution — no feature without a named i2i/u2i/u2u consumer
firstp-prefer-directly-observed-over-learned — observed columns first, learned embeddings second
firstp-reject-features-you-cannot-serve-at-inference — training-serving parity starts at design time
firstp-kill-features-a-popularity-baseline-already-captures — correlation screen before registration
3. Image Feature Extraction (HIGH)
vision-use-clip-for-zero-shot-listing-embeddings — zero-shot CLIP ships in a week
vision-detect-room-types-before-detecting-amenities — room prior conditions the amenity threshold
vision-quantify-image-quality-separately-from-content — blur, lighting, aesthetic as their own features
vision-extract-per-object-counts-not-just-presence — n_bed = 4 beats has_bed = true
vision-pool-embeddings-across-a-listings-photo-set — pooled listing vector; per-photo stored alongside
vision-fine-tune-on-your-domain-when-clip-underperforms — contrastive fine-tune only after zero-shot plateaus
4. Listing Text and Metadata Extraction (HIGH)
listing-declare-categorical-fields-for-bounded-vocabularies — bounded vocab → categorical, validated on write
listing-multi-hot-encode-amenity-lists — fixed amenity vocabulary → multi-hot vector
listing-hash-geo-to-hierarchies-not-raw-lat-lon — H3 at multiple resolutions
listing-embed-description-with-pretrained-sentence-encoder — all-MiniLM-L6-v2 for cheap semantic text features
listing-extract-stay-duration-shape-not-just-length — bin + holiday overlap + flexibility, not raw day count
listing-encode-pet-requirements-as-structured-triples — (species, count, special_needs) triples plus free text alongside
5. Sitter Wizard and Profile Extraction (HIGH)
wizard-order-questions-by-information-gain — discriminative questions first, narrative last
wizard-prefer-multiple-choice-over-free-text — categorical features by construction
wizard-make-skips-genuine-and-log-them — skip is signal; defaults destroy it
wizard-capture-experience-as-counts-and-dates — numbers, not adjectives; platform history overrides self-declaration
wizard-separate-hard-constraints-from-soft-preferences — filters vs ranking features
6. Derived Similarity and Affinity (MEDIUM-HIGH)
derive-precompute-i2i-nearest-neighbours-offline — ANN shelf built nightly, served from KV in <5ms
derive-fuse-modalities-before-item-similarity — vision + text + structured, weighted and normalised
derive-use-two-tower-for-user-item-affinity — dual encoder trained on interactions; ANN-retrieval-ready
derive-score-u2u-as-symmetric-mutual-fit — min(P(owner), P(sitter)); one-sided scoring produces wasted requests
derive-decompose-affinity-into-interpretable-subscores — fit/safety/logistics/price subscores + blend
derive-cache-user-embedding-with-short-ttl — session-level cache, 60-300s TTL
7. Feature Quality and Governance (MEDIUM-HIGH)
quality-version-feature-definitions-in-one-registry — one name, one implementation, one owner
quality-serve-training-and-inference-from-one-store — feature store as the single source of truth
quality-gate-features-on-coverage-and-drift — coverage floor + PSI alarm
quality-scrub-pii-before-features-leave-secure-zone — face blur and regex scrubbing before encoding
quality-freeze-feature-schemas-per-model-version — schema hash pinned to model artifact
8. Incremental Rollout and Value Proof (MEDIUM)
prove-ship-one-feature-at-a-time — one feature, one experiment, one decision
prove-measure-lift-against-feature-ablated-variant — ablation isolates the feature from incidental changes
prove-kill-features-that-dont-earn-maintenance — quarterly kill review on attributed lift
prove-dedicate-random-exploration-slice-to-new-features — 3-5% slice catches offline-close-to-tied winners
prove-retain-feature-free-baseline-permanently — popularity baseline as drift anchor
Discovering New Features
One playbook composes the rules into an end-to-end workflow:
references/playbooks/discovering.md — Discover new features from raw marketplace assets: a seven-step workflow that starts with an asset audit and a decision decomposition and ends with a shipped ablation A/B against a feature-ablated baseline. Use when the task is "what should we build next?" rather than "fix this specific feature."
Read the playbook first when the task is an open-ended "how do we extract more signal from X?" Read individual rules when a specific implementation question arises.
How to Use
- Read
references/_sections.md for category structure and cascade rationale
- Read
gotchas.md for accumulated diagnostic lessons before suggesting interventions
- Read
references/playbooks/discovering.md to plan a new feature discovery cycle
- Read individual rule files under
references/ when a specific task matches the rule title
- Use
assets/templates/_template.md to author new rules as the skill grows
Related Skills
marketplace-personalisation — Post-extraction personalisation on AWS Personalize: event tracking, schema design, two-sided matching, cold start, feedback loops. Hand off once your features are in the store and you are ready to train a ranker.
marketplace-search-recsys-planning — OpenSearch retrieval planning: query understanding, index design, ranking, search-plus-recs blending. Hand off when the bottleneck is retrieval rather than feature availability.
marketplace-pre-member-personalisation — Pre-member journey from anonymous visit to paid membership: anonymous signal inference, onboarding intent capture, pre-member measurement. Hand off at the paid-member boundary.
Reference Files
| File |
Description |
| references/_sections.md |
Category definitions, impact ordering, cascade rationale |
| references/playbooks/discovering.md |
End-to-end feature discovery playbook |
| gotchas.md |
Accumulated feature-engineering diagnostic lessons (living) |
| assets/templates/_template.md |
Template for authoring new rules |
| metadata.json |
Version, discipline, authoritative references |
1---2name: marketplace-recsys-feature-engineering-23description: Use this skill whenever deciding what features to extract from raw marketplace assets — listing photos, owner-entered listing metadata, sitter wizard responses — to power item-to-item (similar listings), user-to-item (homefeed ranking), or user-to-user (mutual-fit matching) recommenders in a two-sided trust marketplace. Covers asset auditing, first-principles feature decomposition from the decision the user is making, vision-feature extraction (CLIP, room-type classification, amenity detection, aesthetic and quality scoring), listing text and metadata encoding (categoricals, multi-hot amenities, H3 geo-hashing, sentence-transformer description embeddings, structured pet triples), sitter wizard design (information-gain ordering, multiple-choice over free text, genuine skippability, hard constraint versus soft preference), derived-composition patterns for i2i / u2i / u2u (precomputed ANN shelves, multi-modal fusion, two-tower affinity, symmetric mutual-fit scoring, interpretable subscores), feature quality gove4---56# Marketplace Engineering Recsys Feature Engineering Best Practices78Comprehensive first-principles guide for deriving usable recommender features from the raw assets of a two-sided trust marketplace — listing photos, owner-supplied listing metadata, and sitter wizard responses — for item-to-item, user-to-item, and user-to-user solutions. Contains 44 rules across 8 categories ordered by cascade impact on the feature-engineering lifecycle, plus one playbook that composes the rules into an end-to-end feature discovery workflow.910This skill is the **upstream precursor** to `marketplace-personalisation` (AWS Personalize) and `marketplace-search-recsys-planning` (OpenSearch retrieval). Those skills treat features as inputs they already have; this skill is about deciding what features to *build* from the raw assets, which decisions they serve, and how to prove each one is worth its maintenance cost.1112## When to Apply1314Reference this skill when:1516- Planning what to extract from listing photos, descriptions, or amenity lists to power i2i similarity or u2i ranking17- Designing or revising the sitter onboarding wizard with recsys features as the primary output18- Deciding whether to build a vision embedding pipeline, a text encoder, or neither — and in what order19- Composing existing base features into item-to-item, user-to-item, or user-to-user scoring20- Auditing an existing feature store for coverage, drift, PII, duplication, or orphan features21- Choosing a ship/kill criterion for a new recsys feature and designing the ablation A/B test22- Answering the question: "we want to improve the similar-homes shelf — what feature should we build?"2324## Setup2526This skill has no user-specific configuration — it is self-contained. References are live URLs to engineering blogs from Airbnb, Pinterest, DoorDash, Uber, Netflix, and Google, to open-source libraries (Feast, Sentence-Transformers, Hugging Face CLIP, H3), to foundational academic papers (Airbnb KDD 2018, Pinterest ItemSage, YouTube Semantic IDs, PinSage), and to Google's Rules of Machine Learning.2728## Rule Categories2930Categories are ordered by cascade impact on the feature-engineering lifecycle: auditing mistakes build features on data that does not exist, first-principles mistakes produce features that do not map to real decisions, extraction mistakes poison everything downstream, and so on. Fix earlier-stage problems before later-stage problems.3132| # | Category | Prefix | Impact |33|---|----------|--------|--------|34| 1 | Asset Audit and Inventory | `audit-` | CRITICAL |35| 2 | First-Principles Feature Decomposition | `firstp-` | CRITICAL |36| 3 | Image Feature Extraction | `vision-` | HIGH |37| 4 | Listing Text and Metadata Extraction | `listing-` | HIGH |38| 5 | Sitter Wizard and Profile Extraction | `wizard-` | HIGH |39| 6 | Derived Similarity and Affinity | `derive-` | MEDIUM-HIGH |40| 7 | Feature Quality and Governance | `quality-` | MEDIUM-HIGH |41| 8 | Incremental Rollout and Value Proof | `prove-` | MEDIUM |4243## Quick Reference4445### 1. Asset Audit and Inventory (CRITICAL)4647- [`audit-measure-coverage-before-modelling`](references/audit-measure-coverage-before-modelling.md) — reject fields below 80% coverage from the feature plan48- [`audit-sample-every-asset-type-end-to-end`](references/audit-sample-every-asset-type-end-to-end.md) — pull 100 real instances through the real fetch path before planning49- [`audit-verify-rights-and-privacy-before-extraction`](references/audit-verify-rights-and-privacy-before-extraction.md) — ToS, GDPR, consent, face blur before encoding50- [`audit-quantify-freshness-per-asset`](references/audit-quantify-freshness-per-asset.md) — age distribution + expiry + refresh bucket51- [`audit-separate-raw-assets-from-derived-features`](references/audit-separate-raw-assets-from-derived-features.md) — raw immutable in object store, derived versioned in feature store5253### 2. First-Principles Feature Decomposition (CRITICAL)5455- [`firstp-start-from-the-decision-not-the-algorithm`](references/firstp-start-from-the-decision-not-the-algorithm.md) — decision first, sub-judgments second, tools last56- [`firstp-ask-what-signal-a-human-uses`](references/firstp-ask-what-signal-a-human-uses.md) — interview 8-12 owners and sitters; features trace back to quotes57- [`firstp-tie-every-feature-to-a-specific-solution`](references/firstp-tie-every-feature-to-a-specific-solution.md) — no feature without a named i2i/u2i/u2u consumer58- [`firstp-prefer-directly-observed-over-learned`](references/firstp-prefer-directly-observed-over-learned.md) — observed columns first, learned embeddings second59- [`firstp-reject-features-you-cannot-serve-at-inference`](references/firstp-reject-features-you-cannot-serve-at-inference.md) — training-serving parity starts at design time60- [`firstp-kill-features-a-popularity-baseline-already-captures`](references/firstp-kill-features-a-popularity-baseline-already-captures.md) — correlation screen before registration6162### 3. Image Feature Extraction (HIGH)6364- [`vision-use-clip-for-zero-shot-listing-embeddings`](references/vision-use-clip-for-zero-shot-listing-embeddings.md) — zero-shot CLIP ships in a week65- [`vision-detect-room-types-before-detecting-amenities`](references/vision-detect-room-types-before-detecting-amenities.md) — room prior conditions the amenity threshold66- [`vision-quantify-image-quality-separately-from-content`](references/vision-quantify-image-quality-separately-from-content.md) — blur, lighting, aesthetic as their own features67- [`vision-extract-per-object-counts-not-just-presence`](references/vision-extract-per-object-counts-not-just-presence.md) — `n_bed = 4` beats `has_bed = true`68- [`vision-pool-embeddings-across-a-listings-photo-set`](references/vision-pool-embeddings-across-a-listings-photo-set.md) — pooled listing vector; per-photo stored alongside69- [`vision-fine-tune-on-your-domain-when-clip-underperforms`](references/vision-fine-tune-on-your-domain-when-clip-underperforms.md) — contrastive fine-tune only after zero-shot plateaus7071### 4. Listing Text and Metadata Extraction (HIGH)7273- [`listing-declare-categorical-fields-for-bounded-vocabularies`](references/listing-declare-categorical-fields-for-bounded-vocabularies.md) — bounded vocab → categorical, validated on write74- [`listing-multi-hot-encode-amenity-lists`](references/listing-multi-hot-encode-amenity-lists.md) — fixed amenity vocabulary → multi-hot vector75- [`listing-hash-geo-to-hierarchies-not-raw-lat-lon`](references/listing-hash-geo-to-hierarchies-not-raw-lat-lon.md) — H3 at multiple resolutions76- [`listing-embed-description-with-pretrained-sentence-encoder`](references/listing-embed-description-with-pretrained-sentence-encoder.md) — all-MiniLM-L6-v2 for cheap semantic text features77- [`listing-extract-stay-duration-shape-not-just-length`](references/listing-extract-stay-duration-shape-not-just-length.md) — bin + holiday overlap + flexibility, not raw day count78- [`listing-encode-pet-requirements-as-structured-triples`](references/listing-encode-pet-requirements-as-structured-triples.md) — `(species, count, special_needs)` triples plus free text alongside7980### 5. Sitter Wizard and Profile Extraction (HIGH)8182- [`wizard-order-questions-by-information-gain`](references/wizard-order-questions-by-information-gain.md) — discriminative questions first, narrative last83- [`wizard-prefer-multiple-choice-over-free-text`](references/wizard-prefer-multiple-choice-over-free-text.md) — categorical features by construction84- [`wizard-make-skips-genuine-and-log-them`](references/wizard-make-skips-genuine-and-log-them.md) — skip is signal; defaults destroy it85- [`wizard-capture-experience-as-counts-and-dates`](references/wizard-capture-experience-as-counts-and-dates.md) — numbers, not adjectives; platform history overrides self-declaration86- [`wizard-separate-hard-constraints-from-soft-preferences`](references/wizard-separate-hard-constraints-from-soft-preferences.md) — filters vs ranking features8788### 6. Derived Similarity and Affinity (MEDIUM-HIGH)8990- [`derive-precompute-i2i-nearest-neighbours-offline`](references/derive-precompute-i2i-nearest-neighbours-offline.md) — ANN shelf built nightly, served from KV in <5ms91- [`derive-fuse-modalities-before-item-similarity`](references/derive-fuse-modalities-before-item-similarity.md) — vision + text + structured, weighted and normalised92- [`derive-use-two-tower-for-user-item-affinity`](references/derive-use-two-tower-for-user-item-affinity.md) — dual encoder trained on interactions; ANN-retrieval-ready93- [`derive-score-u2u-as-symmetric-mutual-fit`](references/derive-score-u2u-as-symmetric-mutual-fit.md) — `min(P(owner), P(sitter))`; one-sided scoring produces wasted requests94- [`derive-decompose-affinity-into-interpretable-subscores`](references/derive-decompose-affinity-into-interpretable-subscores.md) — fit/safety/logistics/price subscores + blend95- [`derive-cache-user-embedding-with-short-ttl`](references/derive-cache-user-embedding-with-short-ttl.md) — session-level cache, 60-300s TTL9697### 7. Feature Quality and Governance (MEDIUM-HIGH)9899- [`quality-version-feature-definitions-in-one-registry`](references/quality-version-feature-definitions-in-one-registry.md) — one name, one implementation, one owner100- [`quality-serve-training-and-inference-from-one-store`](references/quality-serve-training-and-inference-from-one-store.md) — feature store as the single source of truth101- [`quality-gate-features-on-coverage-and-drift`](references/quality-gate-features-on-coverage-and-drift.md) — coverage floor + PSI alarm102- [`quality-scrub-pii-before-features-leave-secure-zone`](references/quality-scrub-pii-before-features-leave-secure-zone.md) — face blur and regex scrubbing before encoding103- [`quality-freeze-feature-schemas-per-model-version`](references/quality-freeze-feature-schemas-per-model-version.md) — schema hash pinned to model artifact104105### 8. Incremental Rollout and Value Proof (MEDIUM)106107- [`prove-ship-one-feature-at-a-time`](references/prove-ship-one-feature-at-a-time.md) — one feature, one experiment, one decision108- [`prove-measure-lift-against-feature-ablated-variant`](references/prove-measure-lift-against-feature-ablated-variant.md) — ablation isolates the feature from incidental changes109- [`prove-kill-features-that-dont-earn-maintenance`](references/prove-kill-features-that-dont-earn-maintenance.md) — quarterly kill review on attributed lift110- [`prove-dedicate-random-exploration-slice-to-new-features`](references/prove-dedicate-random-exploration-slice-to-new-features.md) — 3-5% slice catches offline-close-to-tied winners111- [`prove-retain-feature-free-baseline-permanently`](references/prove-retain-feature-free-baseline-permanently.md) — popularity baseline as drift anchor112113## Discovering New Features114115One playbook composes the rules into an end-to-end workflow:116117- [`references/playbooks/discovering.md`](references/playbooks/discovering.md) — Discover new features from raw marketplace assets: a seven-step workflow that starts with an asset audit and a decision decomposition and ends with a shipped ablation A/B against a feature-ablated baseline. Use when the task is "what should we build next?" rather than "fix this specific feature."118119Read the playbook first when the task is an open-ended "how do we extract more signal from X?" Read individual rules when a specific implementation question arises.120121## How to Use122123- Read [`references/_sections.md`](references/_sections.md) for category structure and cascade rationale124- Read [`gotchas.md`](gotchas.md) for accumulated diagnostic lessons before suggesting interventions125- Read [`references/playbooks/discovering.md`](references/playbooks/discovering.md) to plan a new feature discovery cycle126- Read individual rule files under `references/` when a specific task matches the rule title127- Use [`assets/templates/_template.md`](assets/templates/_template.md) to author new rules as the skill grows128129## Related Skills130131- **`marketplace-personalisation`** — Post-extraction personalisation on AWS Personalize: event tracking, schema design, two-sided matching, cold start, feedback loops. Hand off once your features are in the store and you are ready to train a ranker.132- **`marketplace-search-recsys-planning`** — OpenSearch retrieval planning: query understanding, index design, ranking, search-plus-recs blending. Hand off when the bottleneck is retrieval rather than feature availability.133- **`marketplace-pre-member-personalisation`** — Pre-member journey from anonymous visit to paid membership: anonymous signal inference, onboarding intent capture, pre-member measurement. Hand off at the paid-member boundary.134135## Reference Files136137| File | Description |138|------|-------------|139| [references/_sections.md](references/_sections.md) | Category definitions, impact ordering, cascade rationale |140| [references/playbooks/discovering.md](references/playbooks/discovering.md) | End-to-end feature discovery playbook |141| [gotchas.md](gotchas.md) | Accumulated feature-engineering diagnostic lessons (living) |142| [assets/templates/_template.md](assets/templates/_template.md) | Template for authoring new rules |143| [metadata.json](metadata.json) | Version, discipline, authoritative references |