Schedule List ViewGrid View

Machine learning (ML) algorithms are good at learning new behaviors but bad at identifying when those behaviors are harmful or don’t make sense. Bias, ethics, and fairness are big risk factors in ML. However, we creators have a lot of experience dealing with intelligent beings—one another. Jerry Overton uses this common sense to build a checklist for protecting against ethical violations with ML.

The success of deep learning has reached the realm of structured data in the past few years, where neural networks have been shown to improve the effectiveness and predictability of recommendation engines. Oliver Gindele offers a brief overview of such deep recommender systems and explains how they can be implemented in TensorFlow.

When emergency events occur, social signals and sensor data are generated. Alex Jaimes explains how to apply machine learning and deep learning to process large amounts of heterogeneous data from various sources in real time, with a particular focus on how such information can be used for emergencies and in critical events for first responders and for other social good use cases.

Fast data and stream processing are essential for making Lyft rides a good experience for passengers and drivers. Lyft's systems need to track and react to event streams in real time to update locations, compute routes and estimates, balance prices, and more. Thomas Weise offers an overview of the streaming platform that powers these use cases.

How do you leverage the flexibility and extreme scale of the public cloud and the Apache Kafka ecosystem to build scalable, mission-critical machine learning infrastructures that span multiple public clouds—or bridge your on-premises data center to the cloud? Join Kai Wähner to learn how to use technologies such as TensorFlow with Kafka’s open source ecosystem for machine learning infrastructures.

In the Kubernetes world, where declarative resources are a first-class citizen, running complicated workloads across distributed infrastructure is easy, and processing big data workloads using Spark is common practice, we can finally look at constructing a hybrid system of running Spark in a distributed cloud native way. Join respective experts Kris Nova and Holden Karau for a fun adventure.

Time series databases require ingesting high volumes of structured data, answering complex, performant queries for recent and historical time intervals, and performing specialized time-centric analysis and data management. Michael Freedman explains how to avoid these operational problems by reengineering Postgres to serve as a general data platform, including high-volume time series workloads.

Apache Impala is an MPP SQL query engine for planet-scale queries. When set up and used properly, Impala is able to handle hundreds of nodes and tens of thousands of queries hourly. Manish Maheshwari explains how to avoid pitfalls in Impala configuration (memory limits, admission pools, metadata management, statistics), along with best practices and anti-patterns for end users or BI applications.

David Josephsen tells the story of how Sparkpost's reliability engineering team abandoned ELK for a DIY schema-on-read logging infrastructure. Join in to learn the architectural details, trials, and tribulations from the company's Internal Event Hose data ingestion pipeline project, which uses Fluentd, Kinesis, Parquet, and AWS Athena to make logging sane.

Elliot West and Jay Green share approaches for applying software engineering best practices to SQL-based data applications to improve maintainability and data quality. Using open source tools, Elliot and Jay show how to build effective test suites for Apache Hive code bases and offer an overview of Mutant Swarm, a tool to identify weaknesses in tests and to measure SQL code coverage.

Data warehouses have traditionally run in the data center, and in recent years, they've been adapted to be more cloud native. Greg Rahn discusses a number of emerging trends and technologies that will impact how data warehouses are run both in the cloud and on-premises and explains what that means for architects, administrators, and end users.

Organizations need to keep ahead of their competition by using the latest AI, ML, and DL technologies such as Spark, TensorFlow, and H2O. The challenge is in how to deploy these tools and keep them running in a consistent manner while maximizing the use of scarce hardware resources, such as GPUs. Thomas Phelan discusses the effective deployment of such applications in a container environment.

Jian Zhang, Chendi Xue, and Yuan Zhou explore the challenges of migrating big data analytics workloads to the public cloud (e.g., performance lost and missing features) and demonstrate how to use a new in-memory data accelerator leveraging persistent memory and RDMA NICs to resolve this issues and enable new opportunities for big data workloads on the cloud.

Travel platform Expedia Group likes to give its data teams flexibility and autonomy to work with different technologies. However, this approach generates challenges that cannot be solved by existing tools. Pradeep Bhadani and Elliot West explain how the company built a unified virtual data lake on top of its many heterogeneous and distributed data platforms.

Analysts and researchers studying Wikipedia are hungry for long-term data to build experiments and feed data-driven decisions. But Wikipedia has a strict privacy policy that prevents storing privacy-sensitive data over 90 days. Marcel Ruiz Forns explains how the Wikimedia Foundation's analytics team is working on a vegan data diet to satisfy both.

Enterprise data on customers, vendors, and products is often siloed and represented differently in diverse systems, hurting analytics, compliance, regulatory reporting, and 360 views. Traditional rule-based MDM systems with legacy architectures struggle to unify this growing data. Sonal Goyal offers an overview of a modern master data application using Spark, Cassandra, ML, and Elastic.

Apache Oozie and Apache Airflow (incubating) are both widely used workflow orchestration systems, the former focusing on Apache Hadoop jobs. Feng Lu, James Malone, Apurva Desai, and Cameron Moberg explore an open source Oozie-to-Airflow migration tool developed at Google as a part of creating an effective cross-cloud and cross-system solution.

Teams today rely on dictionaries of collective wisdom—a mixed bag with regard to correctness: some datasets have accurate attribute details, while others are incorrect and outdated. This significantly impacts productivity of analysts and scientists. Sandeep Uttamchandani outlines three patterns to better manage data dictionaries.

Knowledge of customers' location and travel patterns is important for many companies, including German telco service operator Deutsche Telekom. Václav Surovec and Gabor Kotalik explain how a commercial roaming project using Cloudera Hadoop helped the company better analyze the behavior of its customers from 10 countries and provide better predictions and visualizations for management.

Features are key to driving impact with AI at all scales, allowing organizations to dramatically accelerate innovation and time to market. Willem Pienaar and Zhiling Chen explain how GOJEK, Indonesia's first billion-dollar startup, unlocked insights in AI by building a feature store called Feast, and the lessons they learned along the way.

In upstream oil and gas, a vast amount of the data requested for analytics projects is scientific data: physical measurements about the real world. Historically, this data has been managed library style, but a new system was needed to best provide this data. Sun Maria Lehmann and Jane McConnell share architectural best practices learned from their work with subsurface data.

Max Schultze details Zalondo's end-to-end data integration platform to serve analytical use cases and machine learning throughout the company, covering raw data collection, standardized data preparation (binary conversion, partitioning, etc.), user-driven analytics, and machine learning.

Eoin O'Flanagan and Darragh McConville explain how NewDay built a high-performance contemporary data processing platform from the ground up on AWS. Join in to explore the company's journey from a traditional legacy onsite data estate to an entirely cloud-based PCI DSS-compliant platform.

Artificial intelligence and machine learning are now popularly used terms, but how do you make use of these techniques without throwing away the valuable knowledge of experienced employees? Rebecca Simmonds and Michael McCune delve into this idea with examples of how distributed machine learning frameworks fit together naturally with business rules management systems.

Cloudflare powers nearly 10 percent of all Internet requests worldwide, absorbing some of the largest DDoS attacks. Learn how we use ClickHouse and SQL to simplify our data pipelines on a global scale while experiencing over 10 million events per second.

The Embulk data migration tool offers a convenient way to load data in to a variety of systems with basic configuration. Jason Bell offers an overview of the Embulk tool and outlines some common data migration scenarios that a data engineer could employ using the tool.

Fabio Ferraretto and Claudia Regina Laselva explain how Hospital Albert Einstein and Accenture evolved patient flow experience and efficiency with the use of applied AI, statistics, and combinatorial math, allowing the hospital to anticipate E2E visibility within patient flow operations, from admission of emergency and elective demands to assignment and medical releases.

Not surprisingly, there's no single approach to embracing data-driven innovations within any industry vertical. However, some enterprises are doing a better job than others when it comes to establishing a culture, process, and infrastructure that lends itself to data-driven innovations. Vidya Raman explores some key foundational ingredients that span multiple industries.

Rosaria Silipo shares a collection of past data science projects. While the structure is often similar—data collection, data transformation, model training, deployment—each required its own special trick, whether a change in perspective or a particular technique to deal with special case and special business questions.

Robert Cohen discusses the skills that employers are seeking from employees in digital jobs, linked to the new software hierarchy driving digital transformation. Robert describes this software hierarchy as one that ranges from DevOps, CI/CD, and microservices to Kubernetes and Istio. This hierarchy is used to define the jobs that are central to data-driven digital transformation.

Data science is an approachable field given the right framing. Often, though, practitioners and executives are describing opportunities using completely different languages. Brandy Freitas walks you through developing context and vocabulary around data science topics to help build a culture of data within your organization.

Alasdair Allan explains why the current age, where privacy is no longer "a social norm," may not long survive the coming of the internet of things, as new smart embedded hardware may cause the demise of large-scale data harvesting. Smart devices will process data at the edge, allowing us to extract insights from the data without storing potentially privacy- and GDPR-infringing data.

Many companies delay addressing core improvements in increasing revenues, reducing costs and risk exposure by tying changes to a to-be-hired data scientist. Drawing on three customer examples, Jack Norris explains how to achieve excellent results faster by starting with domain experience and helping developers and analysts better leverage data with available and understandable analytics.

Nikki Rouda shares key trends in data lakes and analytics and explains how they shape the services offered by AWS. Specific topics include the rise of machine-generated data and semistructured and unstructured data as dominant sources of new data, the move toward serverless, SPI-centric computing, and the growing need for local access to data from users around the world.

Your team is building machine learning capabilities. Dean Wampler demonstrates how to integrate these capabilities in streaming data pipelines so you can leverage the results quickly and update them as needed and covers challenges such as how to build long-running services that are very reliable and scalable and how to combine a spectrum of very different tools, from data science to operations.

David Dogon dives into a best practice use case for detecting fraud at a financial institution and details a dynamic and robust monitoring system that successfully detects unwanted client behavior. Join in to learn how machine learning models can provide a solution in cases where traditional systems fall short.

Moshe Wasserblat offers an overview of NLP Architect, an open source DL NLP library that provides SOTA NLP models, making it easy for researchers to implement NLP algorithms and for data scientists to build NLP-based solutions for extracting insight from textual data to improve business operations.

Graph query language is the key to unleash the value from connected data. Mingxi Wu outlines the eight prerequisites of a practical graph query language, drawn from six years' experience dealing with real-world graph analytical use cases. Along the way, Mingxi compares GSQL, Gremlin, Cypher, and SPARQL, pointing out their respective pros and cons.

Supervised machine learning requires large labeled datasets—a prohibitive limitation in many real-world applications. What if machines could learn with fewer labeled examples? Shioulin Sam shares an algorithmic solution that relies on collaboration between humans and machines to label smartly and discusses product possibilities.

Cybersecurity analysts are under siege to keep pace with the ever-changing threat landscape. The analysts are overworked as they are bombarded with and burned out by the sheer number of alerts that they must carefully investigate. Brennan Lodge and Jay Kesavan explain how to use a data science model for alert evaluations to empower your cybersecurity analysts.

Identifying relevant documents quickly and efficiently enhances both user experience and business revenue every day. Sophie Watson demonstrates how to implement learning-to-rank algorithms and provides you with the information you need to implement your own successful ranking system.

Seonmin Kim offers an introduction to activities that mitigate the risk of mobile payments through various data analytical skills, drawn from actual case studies of mobile frauds, along with tree-based machine learning, graph analytics, and statistical approaches.

Reinforcement learning (RL) learns complex processes autonomously like walking, beating the world champion in Go, or flying a helicopter. No big datasets with the “right” answers are needed: the algorithms learn by experimenting. Christian Hidber shows how and why RL works and demonstrates how to apply it to an industrial hydraulics application with 7,000 clients in 42 countries.

Christopher Hooi offers an overview of the Fusion Analytics for Public Transport Event Response (FASTER) system, a real-time advanced analytics solution for early warning of potential train incidents. FASTER uses engineering and commuter-centric IoT data sources to activate contingency plans at the earliest possible time and reduce impact to commuters.

GRDF helps bring natural gas to nearly 11 million customers every day. Alexandre Hubert explains how, in partnership with GRDF, Dataiku worked to optimize the manual process of qualifying addresses to visit and ultimately save GRDF time and money. This solution was the culmination of a yearlong adventure in the land of maintenance experts, legacy IT systems, and Agile development.

Modern deep learning systems allow us to build speech synthesis systems with the naturalness of a human speaker. While there are myriad benevolent applications, this also ushers in a new era of fake news. Scott Stevenson explores the danger of such systems and details how deep learning can also be used to build countermeasures to protect against political disinformation.

In this auditory world, the human brain processes and reacts effortlessly to a variety of sounds. While many of us take this for granted, there are over 360 million in this world who are deaf or hard of hearing. Swetha Machanavajhala and Xiaoyong Zhu explain how to make the auditory world inclusive and meet the great demand in other sectors by applying deep learning on audio in Azure.

Deep learning has enabled massive breakthroughs in offbeat tracks and has enabled better understanding of how an artist paints, how an artist composes music, and so on. Nischal Harohalli Padmanabha and Raghotham Sripadraj discuss their project Deep Learning for Humans and their plans to build a font classifier.

Technological advancements are transforming customer experience, and businesses are beginning to benefit from deep learning innovations to automate call center routing to the most proper agent. Tal Doron explains how to run deep learning models with Intel BigDL and Spark frameworks colocated on an in-memory computing platform to enhance the customer experience without the need for GPUs

Join Ben Lackey to learn how Oracle Cloud Infrastructure's architecture makes it the right place to run compute-intensive partner applications like H20.ai, Cloudera, DataStax, and more.

Program chairs Ben Lorica, Doug Cutting, and Alistair Croll welcome you to the second day of keynotes.

Chris Taggart explains the benefits of white box data and outlines the structural shifts that are moving the data world toward it.

Shingai Manjengwa shares insights from teaching data science to 300,000 online learners, second-career college graduates, and grade 12/6th form high school students, explaining how business leaders can increase data science skill sets across different levels and functions in an organization to create real and measurable value from data.

Companies that harness creativity and data in tandem have growth rates twice as high as companies that don’t. David Boyle shares lessons from his successes and failures in trying to do just that across presidential politics, with pop stars, and with power brands in the world of luxury goods. Join in to find out how analysts can work differently to build these partnerships and unlock this growth.

BMW Group is an extraordinary company. As a technology pioneer it's an enterprise that recognizes the value that data to offers to the business. The company's global platform draws data from over 150 different systems and delivers governed data to various divisions. Join Amr Awadallah and Tobias Burger to discover some of BMW's most important use cases leveraging data from the edge to AI.

Ogilvy's Mike Tidmarsh looks at how data and AI are radically reshaping the world of marketing communications and explores the impacts—good and bad—for professionals and consumers alike.

Big data analytics and AI draw nonintuitive and unverifiable inferences about the behaviors, preferences, and lives of individuals. These inferences draw on diverse and feature-rich data of unpredictable value and create new opportunities for discriminatory, biased, and invasive decision making. Sandra Wachter discusses how this expands potential victims of discrimination and potential harm.

Topic Table discussions are a great way to informally network with people in similar industries or interested in the same topics.

Gather before keynotes on Thursday morning for a speed networking event. Enjoy casual conversation while meeting fellow attendees. If one of your goals at Strata is to meet new people, this session will jumpstart your networking with fellow attendees.