Catalogue
- Cards
- 165
- Records
- 486
- Workloads
- 19
- Corpus version
- 74847648e25a
197 of 486 records match; the first 20 are shown.
| Kind | Record | Organisation | Year | Setup |
|---|---|---|---|---|
| Decision | Partition catalog clustering to per-shop tasks and filter standalone stores deterministically via Liquid themes before calling LLMs | Shopify | 2026 | |
| Decision | Replace naive linkage-tree clustering with FAISS HNSW plus sparse UPGMA average linkage to form coherent chunks up to 200 products | Shopify | 2026 | |
| Decision | Use a two-stage propose-and-critique LLM flow with dynamic per-chunk Structured Output schemas and normalized product IDs | Shopify | 2026 | |
| Decision | Own the webhook pipeline after customers could not see what happened to a webhook, leaving a hosted queue | Marble | 2026 | |
| Decision | Workers atomically claim a delivery before sending so overlapping workers or retried queue messages do not send duplicates | Marble | 2026 | |
| Decision | Replace customer-owned social-graph ingestion pipelines with a self-managed data warehouse ingestion service | Meta | 2026 | |
| Decision | Migrate jobs through three gated phases—shadow, reverse shadow and cleanup—verified on row counts, checksums, latency and resources | Meta | 2026 | |
| Decision | Contain CDC bad-data propagation via paired reverse-shadow backfills and metadata bad-partition flags | Meta | 2026 | |
| Decision | Batch shadow migrations under limited capacity, holding back jobs with known bugs and reusing legacy snapshots | Meta | 2026 | |
| Decision | Automate lifecycle promotion from Scuba signals and run hourly mismatch row-level debugging | Meta | 2026 | |
| Failure | One customer's endpoint outage made deliveries time out, filled the shared worker pool and delayed Directory Sync webhooks for everyone | WorkOS | 2025 | |
| Decision | Split feature engineering into distributed PySpark pre-processing on Databricks and vertical Python transformations on SageMaker | Zalando | 2025 | |
| Decision | Use LightGBM with Nixtla MLForecast to combine demand forecast training and inference in one SageMaker job | Zalando | 2025 | |
| Decision | Use dual offline and online SageMaker Feature Store tiers and write interactive updates back to the offline store | Zalando | 2025 | |
| Decision | Serve interactive partner inventory-setting updates asynchronously through SQS, Lambda and event-stream notifications | Zalando | 2025 | |
| Decision | Isolate every pipeline run on dedicated transient Databricks job clusters and SageMaker jobs via zFlow | Zalando | 2025 | |
| Failure | Search compaction invalidations triggered a WebSocket reconnect storm that was prolonged by a manual deploy onto default free-tier hardware | Convex | 2025 | |
| Decision | Move ad tracking metadata out of client opaque tokens into a transient key-value Ads Metadata Registry | Netflix | 2025 | |
| Decision | Centralize ad event ingestion behind Ads Event Publisher with a unified contract and place all consumers and sessionization downstream | Netflix | 2025 | |
| Decision | Combine real-time Flink-to-Druid ad metrics with offline correction and separate offline billing workflows | Netflix | 2025 |
177 more records. Sign in to see all 197.