
Bert Latamore13
























#IBM #Spark Tech Center Principal Engineer Nick Pentreath live now on #theCUBE from Spark Summit East @DVellante @GGilbert41 http://bit.ly/2kOdAJ...

Bert Latamore
Overarching goal is to drive adoption in enterprise customers & make #Spark enterprise-ready. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
#IBM invests in Open Source technologies that it sees as transformational. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
#IBM backing #Spark as a next generation analaytics platform. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
Before we just dumped data into the data lakes & silos. Now what are we going to do with it. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
Lots of use cases for machine learning inc. recommendation engines. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
Making recommendations to online customers as to what they should buy is a classic. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
Fraud detection in financial services & enterprises are another classic use case. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
No we have good models for these & other use cases are in the Spark library. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
What is missing? A huge complex workflow in the end-to-end story. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
Feed multiple data streams into your database, then do the data science & then deploy the machine learning algorithm. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
it is not magic that you throw machine learning at a database & everybody's happy. U have to fulfill the user's needs. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
It's always difficult to make decisions about timeframes. There is a long way to go. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
Time gap for predicdtions is down to real-time. Need better feedback, monitoring, end-user experience. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
There's a lot of work still to be done. Areas of active research in academic field on improving these systems. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
Improving the quality of fraud detection -- we have a long way to go still. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
The enterprise-applied machine-learning problem has moved from the academic. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
#IBM has announced Watson Machine Learning to productionize the end-to-end machine learning platform. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
When you operate at the scale of #IBM Ur customers will find all the bugs. We try to make it better for everyone. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
we take all the feedback and centralize the fixes in the #Apache #Spark platfoarm. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
We as a community need to decide whether we develop the functionality, adapt open standards or adopt other Open Source projectds. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
Underlying execution engine is on #Spark. Can also run on #Hadoop. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
Millions of lines of code, very powerful engine. Lot of work still to be done for it to be usable in production systems. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
Most of the team is in San Francisco. I'm the only member in Cape Town, working remote. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
Some of the key things I picked up are related to deep learning on #Spark. Interesting work coming from #Intel. @MLNick #theCUBE http://bit.ly/2kOdAJ...

Bert Latamore
Every #SparkSummit there are now projecdts from the community. @MLNick #theCUBE http://bit.ly/2kOdAJ...