Crunch Conference 2019
2019-10-17..2019-10-18
Budapest, Hungary
[Crunch Conference 2019, Crunch Conference 2021, Crunch Conference 2022]
| title | author | length |
|---|---|---|
| Human data interactions: experiments in how humans analyze data | Jeff Leek | 48m |
Group runs experiments inside Coursera courses (over 6 million enrollments) and analyzes code from 35,000 GitHub repositories to measure how humans actually do… | ||
| Infinite* Topic Backlogs with Apache Pulsar | Ivan Kelly | 46m |
Claim: Apache Pulsar scales topic backlogs effectively to “infinite” size by separating serving and storage and using BookKeeper ledgers broken into immutable… | ||
| Building low latency gradient boosted tree predictors on distributed platform | György Móra | 45m |
👉 Learn more about the talk and download the slides at https://crunchconf.com/speaker/GyorgyMora
📬 Sign up to our newsletter so you won't miss the updates… | ||
| Data Lineage with Apache Airflow using Marquez | Willy Lulciuc | 45m |
Marquez is a metadata service that connects to Apache Airflow via a Marquez Airflow library which inspects tasks, parses SQL to auto-register… | ||
| Using adversarial samples to robustify your Neural Network Models | Irina Vidal | 38m |
Shows concrete attacks and defenses: tiny gradient perturbations flip a panda to a gibbon, 3D-printed turtle -> rifle works in the physical world, stickers… | ||
| Make Your Data FABulous | Philipp Krenn | 48m |
Introduces the FAB theory (must choose two of Fast, Accurate, Big) and shows concrete Elasticsearch tradeoffs: term aggregations undercount when shard-level… | ||
| How to Get the Most Out of Your BI Tools | Keshia Rose | 42m |
Shows concrete ways to extend BI: use the web data connector or connector SDK to pull APIs like AlphaVantage, use the Hyper API for local extracts, and use the… | ||
| The data scientist's hierarchy of needs | Benn Stancil | 46m |
A practical hierarchy of needs for data science argues that humans first need clean, usable data and fast SQL access before models or fancy tools matter. Mode… | ||
| The algorithm for precision medicine | Matt Might | 50m |
A parent discovers an ultra‑rare ANGEL1 deficiency by exome sequencing and builds a repeatable pipeline: go from genomic diagnosis to mechanism (overactive… | ||
| Transfer Learning for the Cold Start Problem | Sergey Rubtsovenko | 31m |
A fraud-prevention pipeline for Zalando uses transfer learning and domain adaptation to solve a payment fraud cold-start when a new market has few labels. The… | ||
| Dagster: An open source Python library for building data applications | Nicholas Schrock | 47m |
Dagster is an open source Python library for building ETL, ELT and ML pipelines as graphs of typed computations. It models nodes as solids with inputs, outputs… | ||
| Data as culture as software | Andrea Burbank | 45m |
Six years at Pinterest show how logging and engineering shape usable metrics: raw request logs land in Kafka/Redshift, small derived tables (xd7/xd28) replace… | ||
| Journey of metadata | Shirshanka Das | 46m |
Describes a metadata platform that grows from a crawl-based “warehouse” to a distributed, model-first system powering over a million data assets and serving… | ||
| Stories Beat Statistics: Master the Art & Science of Data Storytelling | Brent Dykes | 51m |
Stories outperform stats: people remember anecdotes (63% recall) far more than numbers (5%), and a charity brochure with one child’s story drives more… | ||
| Toward Human-Centered Machine Learning | Patrick Hall | 43m |
Describes a reproducible workflow that combines GAMs and scalable Bayesian rule lists with post-hoc tools like Shapley values and LIME to produce… | ||
| Beyond Interactive: Empowering Everyone with Jupyter Notebooks | Michelle Ufford | 50m |
Netflix captures 1 trillion events/day into S3, processes mainly with Spark and serves results via Snowflake/Druid/Redshift/Elasticsearch; notebooks become an… | ||
| How to Do Data for Good | Emma Deraze | 52m |
DataKind UK pairs volunteer data people with charities via monthly data dives and longer data cores, then applies practical methods: a regression to impute… | ||
| Clear and Presentation Danger | Andy Cotgreave | 59m |
Practical checklist for presenting data on big screens: zoom or use a remote to highlight specific KPIs, avoid small default fonts (bump labels to readable… | ||
| Recreating the State of the Player | Wesley Kerr | 56m |
Rebuilds complete player profiles by replaying daily event streams, joining game events, store transactions and account snapshots into ordered timelines and… | ||
| The Paradise Papers: Behind the Scenes | Pierre Romera | 40m |
13 million documents (1.4 TB) get ingested, OCRed and indexed so 400 journalists across 95 media can search and share leads; processing uses AWS, a distributed… | ||
| nothing matches | ||