Dinesh Nirmal is Vice President of Development for IBM Data + AI, with global responsibility for building the analytics, data, and AI technology in use by enterprises from governments and international finance firms to startups. Dinesh leads more than a dozen IBM Development Labs around the world. Major releases during his 2-year tenure include IBM Cloud Private for Data—an open, cloud-native information architecture for AI; IBM Integrated Analytics System; Machine Learning for z/OS; and IBM Watson Studio. Products in his portfolio have won major design awards including two Red Dot Awards and the iF Design Award. Additionally, he is the Site Executive for IBM's Silicon Valley Lab.View the profile
About the talk
Dinesh Nirmal is VP for data and AI at IBM, and an expert on emerging technologies. During his keynote at the AI Summit in San Francisco, he proposed an unusual idea: what if we could use augmented and virtual reality to manipulate data, the way we manipulate physical objects?
Welcome to a different TV here at the ASM in San Francisco 2019 be very pleased featuring today by the National Mall. He is the VP of data and AI at IBM Hi, how are you doing? I'm doing great. Thanks. But I look at the street Prince happening one is around data. I mean people really think all that exists in some CSV file on an Enterprise. That's far from the truth database. I load data is fragmented. The quality of data is not good with so many challenges with data at a
tender prices. So I covered that you know, how can you get trusted data to data scientist in a very short. Of time to those that one player that I covered the second one is algorithms because once you develop the model and you deploy it people think okay to work is done. My mom is going well. I'm done but that's not the case. That's where the real work starts because he's the model drifting is the model fair is the Marvel biased all those things you need to make sure the whole Model Management becomes very critical. The third piece was compute because I asked the
proliferation of data happens as more and more data comes in. Enterprise's do you have enough computing power and that's where something like Quantum comes in as part of the data. Peace. I also covered arbr and that's when you mention that's because we have been tracking the data for a very long time mainly through computers. You know, now it's changing to talk into the data you're talking with your dad and walking back. I say that because you know, look at your bar or even home assistant where you can have a conversation with your data. But I'm taking you to the
next level but I'm saying what if you walk with your data or what if you walk into your data where you can have a 3D visualization of the data and your inside the data were you can easily do explainability on all those things easier to take away from my session has more of a futuristic thing in the data, but we are doing some research in there and I want to give a glimpse of that. Intensive up every Stitch of the moment Smith and daddy lap get the trusted
clean data to the data scientist to build model. There's three pillars the data pillar the development of the models and the deployment of the Marvels. There's a lot of research and open-source were going into the development of Tomorrow. Meaning you give it the data runs through a bunch of algorithms and give you the model. But if your Enterprise and the last week is how do I email get the right set of data clean data tractor data quality of the data access to the data all those things becomes very critical. So there's a tremendous
amount of research going into it. For example, how do we order classify machine learning? How do you take the right access back to the data using Optimizer? So Going on within the last last piece is the D your deployment. You know, there's a lot of work going on in there to say, how can we do by C-section? How can you do prediction of by is how can we do the fairness? How can we do it's plain the bloody. So if you ask me for being focused is that we are focused obviously on the middle day of the development, but we
also see that there's a lot of work going on in there but the challenges around that data data cleansing the data quality the data governance, how do we bring a high-end ml in there to help customers get access to trusted data a lot faster and how did he get customers to deploy those models are built a lot more faster using they are that's where we are focusing aggressive you think about the appointment then kind of looking more at the industries and use cases like that. What kind of Industries stands to gain the most intense of this new information revolution? Every
Industry Board Game a tremendous amount by this automation. I'll give you an example for a job right now. You can take a picture of your damaged car and they can you know, obviously using neural net so deep learning they can figure out what part of the car was damaged. What is the level of damage? What is the approximate cost that can predict and then they can do the underwriting right on the on the device itself and redirect you to the closest body shop or within less than an hour, which used to take date. Now order made it to
2 hours. That's one less thing you have airplane fly thousands of devices on the plane, which device is giving you in a trance of data a sense of data, but you can also take that data to say is the device. Function properly vetted mean to be replaced is that what needs to be done? You can connect to your back and Erp system to say this device is, you know wearing and tearing this time to reorder one less place the order in the back and using our Erp. So the whole process that used to
go through days and weeks and months now all of a sudden becomes in an hour show or base when it's about the skills and development skills as an in-house or Alpha thing and this is a real example word. Let's say you are a bank and you have a 30-year veteran who is doing underwriting business. He or she leaves he or she will take 30 years worth of experience with them. I think the workforce is changing more and more building the models by using the training data sets of in all the people who have been doing that work with they know that the money was David them.
There's no leaving before people will leave the knowledge goes with them. But now the models will stay and be the Enterprise for very long time. They can retrain them in a you just have to have different datasets replaying them. So I think the in-house worship hiring I think the whole work force itself is changing to the new paradigm. It's not about your skills. It's about how do I train the model to do in a particular vertical knowledge to get the work done? So I think Enterprises are rethinking the skills, but if I come to a
binary question like that, which one is better, I mean to me personally, it's always good to have in hospital. Because that really helps you to build a core set of skills that always available for you and it's much more better for Enterprise to keep those pills in Houston that Outsource.
Buy this talk
Access to all the recordings of the event
Buy this video
With ConferenceCast.tv, you get access to our library of the world's best conference talks.