{"podcast":{"title":"MLOps Weekly Podcast","slug":"mlops-weekly","podcast_index_feed_id":5487202,"rss_url":"https://media.rss.com/mlops-weekly/feed.xml","website_url":"https://rss.com/podcasts/mlops-weekly","image_url":"https://media.rss.com/mlops-weekly/20220607_010653_39e23ec13c42d0efa239e27bf455ceed.jpg","author":"Simba Khadder","episode_count":32,"summary":"Join each week as we talk to MLOps operators, practitioners, and professionals about the current state of MLOps.","last_synced_at":null,"page_url":"https://stenobird.com/podcast/mlops-weekly"},"episode":{"title":"MLOps Week 28: Featureform's CEO Breaks Down \"Real-Time\" Machine Learning","slug":"mlops-week-28-featureform-s-ceo-breaks-down-real-time-machine-learning","published_at":"2024-03-28T01:16:49+00:00","page_url":"https://stenobird.com/podcast/mlops-weekly/mlops-week-28-featureform-s-ceo-breaks-down-real-time-machine-learning","show_page_url":"https://stenobird.com/podcast/mlops-weekly","url":"https://rss.com/podcasts/mlops-weekly/1410852","audio_url":"https://content.rss.com/episodes/132586/1410852/mlops-weekly/2024_03_28_01_16_05_7fd9f175-e804-4df3-b54e-fad7e9635fb1.mp3","summary":"Stop using 'real-time' as a vague buzzword and start defining it through latency, online serving, and feature freshness. This episode provides a framework for breaking down ML requirements into actionable engineering constraints.","meta_description":"Learn how to define real-time machine learning by analyzing latency budgets, online serving, and the trade-offs between batch, streaming, and on-demand fe…","key_points":["Main idea: Real-time ML is not a single metric but a combination of latency, online serving, and feature freshness","Practical takeaway: Always define a 'latency budget' that accounts for the entire pipeline, including network and feature retrieval, not just model inference","Failure mode: Over-engineering with streaming features when batch processing is sufficient, leading to unnecessary system complexity and higher costs","Main idea: Feature freshness and latency exist in a trade-off; higher freshness typically requires more expensive, higher-latency infrastructure","Practical takeaway: Use on-demand features to combine pre-computed batch/streaming data with request-time data for complex calculations like transaction percentages"],"chapters":[{"start_ms":60000,"title":"Defining the Three Pillars of Real-Time","summary":"An introduction to breaking down real-time ML into latency, online serving, and real-time features."},{"start_ms":145000,"title":"The Latency Budget","summary":"How to evaluate acceptable latency based on use cases like fraud detection versus recommendation engines."},{"start_ms":220000,"title":"Online Serving vs. Offline Models","summary":"Distinguishing between models that must be always-on and those that can run in a batch setting."},{"start_ms":300000,"title":"The Stakes of Online Models","summary":"Understanding the operational importance and immediate impact of models running in an online environment."},{"start_ms":455000,"title":"The Two Axes of Real-Time Features","summary":"Analyzing features through the dual lenses of latency (speed of retrieval) and freshness (data staleness)."},{"start_ms":535000,"title":"Optimizing Feature Windows","summary":"How choosing appropriate time windows for features can reduce compute costs and latency."},{"start_ms":615000,"title":"Batch, Streaming, and On-Demand Features","summary":"A deep dive into the three feature types, their implementation complexities, and when to use each."},{"start_ms":930000,"title":"The Complexity of Streaming","summary":"The engineering challenges and increased failure modes associated with streaming data pipelines."}],"topics":["Machine Learning","MLOps","Real-time Systems","Feature Engineering","Data Latency","Streaming Data","Online Serving","System Architecture"],"duration_seconds":1107,"processing_state":"processed","actions":[{"name":"request_transcript","method":"POST","url":"https://stenobird.com/v1/public/podcasts/mlops-weekly/episodes/mlops-week-28-featureform-s-ceo-breaks-down-real-time-machine-learning/transcription-requests","description":"Idempotently request low-priority transcript generation for this episode."},{"name":"read_markdown","method":"GET","url":"https://stenobird.com/podcast/mlops-weekly/mlops-week-28-featureform-s-ceo-breaks-down-real-time-machine-learning.md","description":"Read the agent-friendly Markdown representation of this episode resource."}]}}