Dave Thomas: NewSQL - The New Universal Query Language for Everything.

Back to the Future! In 2020 many new and legacy data systems, such as Spark, Teradata, Mongo, Cassandra, Splunk, Kafka, Flink, Azure, Google, Oracle…, will support a NewSQL.

We begin with a very brief review of the current state of data engineering practice - NoSQL, Big Data, Streaming, Time Series, BASE versus ACID, Graphs; DSLs, Map Reduce, Functional Programming etc. We look at the challenges presented by the plethora of different data formats, languages/tools and eventual consistency.

What if we didn’t have to deal with the challenges of sharding and eventual consistency? We discuss the recent increased use of new distributed ACID databases such as Google Spanner, AWS AuroraDB, Azure CosmosDB etc.

What if we didn’t need to worry about different languages, APIs for programming different databases? What if we could use one language for both batch and streaming? Very recently, at the ACM SIGMOD 2019 database conference, a new emerging data language was presented which seeks to provide a solution.

We refer to this emerging standard as NewSQL, which extends SQL to deal with both streaming and batch. It removes major limitations in SQL and provides the additional capabilities to SQL needed to deal with the velocity, volume and variety of diverse data sources. In this talk, we describe the evolution and major features of the new language. We conclude with a brief discussion of the impact on data engineering, data science and data consumers.

Agustinus Nalwan: Mystique - The Fight Against Rego Plate Cloning

In the rise of AI technologies, there is no doubt that privacy is a very popular topic which normally revolves around the negative impact of AI to our privacy. However despite the norm, we at Carsales built and deployed an AI tech called Mystique which fights crimes and protects user privacy. Mystique detects a rego plate in a photo and blurs it, protecting our private seller and dealer from rego plate cloning.

Rego plate cloning “involves falsifying a genuine number plate and attaching it to another vehicle which is often the same make, model and colour. The criminal may then use the car with cloned plates to drive on tollways without consequences, drive dangerously or conduct other criminal activity, with the genuine vehicle owner often sought to pay fines or tolls.”

In this talk, we are going to share the story about how we built and delivered Mystique and more importantly what lessons we learned in deploying tech at this scale, which processes 150,000 photos a day and touches many areas of business at Carsales.

Dave Thomas: NewSQL - The New Universal Query Language for Everything.

Back to the Future! In 2020 many new and legacy data systems, such as Spark, Teradata, Mongo, Cassandra, Splunk, Kafka, Flink, Azure, Google, Oracle…, will support a NewSQL.

We begin with a very brief review of the current state of data engineering practice - NoSQL, Big Data, Streaming, Time Series, BASE versus ACID, Graphs; DSLs, Map Reduce, Functional Programming etc. We look at the challenges presented by the plethora of different data formats, languages/tools and eventual consistency.

What if we didn’t have to deal with the challenges of sharding and eventual consistency? We discuss the recent increased use of new distributed ACID databases such as Google Spanner, AWS AuroraDB, Azure CosmosDB etc.

What if we didn’t need to worry about different languages, APIs for programming different databases? What if we could use one language for both batch and streaming? Very recently, at the ACM SIGMOD 2019 database conference, a new emerging data language was presented which seeks to provide a solution.

We refer to this emerging standard as NewSQL, which extends SQL to deal with both streaming and batch. It removes major limitations in SQL and provides the additional capabilities to SQL needed to deal with the velocity, volume and variety of diverse data sources. In this talk, we describe the evolution and major features of the new language. We conclude with a brief discussion of the impact on data engineering, data science and data consumers.

Agustinus Nalwan: Mystique - The Fight Against Rego Plate Cloning

In the rise of AI technologies, there is no doubt that privacy is a very popular topic which normally revolves around the negative impact of AI to our privacy. However despite the norm, we at Carsales built and deployed an AI tech called Mystique which fights crimes and protects user privacy. Mystique detects a rego plate in a photo and blurs it, protecting our private seller and dealer from rego plate cloning.

Rego plate cloning “involves falsifying a genuine number plate and attaching it to another vehicle which is often the same make, model and colour. The criminal may then use the car with cloned plates to drive on tollways without consequences, drive dangerously or conduct other criminal activity, with the genuine vehicle owner often sought to pay fines or tolls.”

In this talk, we are going to share the story about how we built and delivered Mystique and more importantly what lessons we learned in deploying tech at this scale, which processes 150,000 photos a day and touches many areas of business at Carsales.

No doubt that AI is the new kids on the block. From as simple as classifying hot-dog vs not hot-dog, recognising flower species and going towards science fiction realm in generating fake videos.

This talk will cover the problem with supervised learning which is what most of current AI technologies are based on and what is the promising trend towards the future of AI with unsupervised learning. As a use case, we will cover how image generation techniques such as Variational Auto-encoder extract knowledge from images in an unsupervised manner.

You must have heard it a few times that AI has beaten human in image recognition. Is that true? Have you seen it yourself? I am going to demonstrate Cyclops, an image recognition we built to recognise car model far better than any human.

From here on, this talk will take you through our journey, how it's all began, why we built the early version of Cyclops and what was the outcome. Furthermore, how we used this technology to dramatically improve consumer experience and built many consumers facing products which we thought was not possible before.

I will then dive down into technical details, starting from how we built Cyclops 1.0 with Tensorflow and how we overcame the training complexity with transfer learning. However, transfer learning comes with a limitation of directional invariance in which I will show what it is and how we overcame it with our novel solution.

Next, I will show you that building a car recognition as complex as Cyclops 2.0 requires a more superior model and modification of our existing transfer learning technique. I will also take you to see problems we faced with low coverage when we are going deeper and how we solved them. I will then investigate how a distributed training can speed up the training process to make this practical.