TDWI | Training & Research | Business Intelligence, Analytics, Big Data, Data Warehousing

5 Minutes with a Chief Analytics Officer: Michael O'Connell of TIBCO Software

O'Connell spoke recently about his work "at the confluence of analytics software and customers' business problems" and what he sees as the future of AI solutions.

• By James E. Powell
• March 10, 2017

Michael O'Connell, Ph.D., is the chief analytics officer at TIBCO Software, where he helps clients with analytics software applications that drive business value. He has written more than 50 scientific papers and several software packages on statistical methods.

He recently spoke with us about his work and what he sees as the future of AI.

Upside: Are you working on anything super interesting right now?

Michael O'Connell: I've been doing a lot of anomaly detection lately -- setting up analytics software systems to detect equipment outages or degradation of service; or to identify financial crime, fraud, or money laundering. Both sensor data and financial transaction data can have many dimensions. Finding meaningful outliers in such multivariate data is a really interesting problem, and we've been developing some simple, useful methods.

My work comes at the confluence of analytics software and our customers' business problems. It's a process to understand the customers' problems and resources and then determine how our knowledge and available products can help them find a solution.

We're infusing our products with AI smarts that help our customers rapidly identify insights and convert these insights into action. Think of personal assistants for "data detective" and "business entrepreneur" personas -- that's what our AI-enabled products and solutions strive to provide.

Addressing customer problems and employing AI and machine learning in solutions are my two biggest passions. Business value comes from providing a smart software user experience to find insights and enabling the user to deploy those insights to action in business processes.

What's your favorite part of your job? Your least favorite part?

Meeting with customers is truly inspiring and gives me the passion and motivation to deliver the best results from the data and problem at hand. I always take the time to listen to our customers and think of a way to solve their problems with the resources available.

I have a set of great products to work with -- our products and the broader data science ecosystem. I enjoy helping our customers identify the most effective solution for their needs. If it's uncharted territory, then taking the time to research a new solution approach can be fun and rewarding.

My least favorite part is probably wrangling the data sources and dealing with the IT systems -- particularly some of the big data systems and open source software components. Thankfully our products are pretty smart and configurable when it comes to data access, prep, and wrangling, and the data science ecosystem of tools is getting stronger every day.

What's a personality trait you think people need to succeed at your job?

Curiosity and a sense of wonder about how the world works -- along with the ability to work hard and fast and put in the 10,000 hours that Malcolm Gladwell talks about. Although it's good to have training in data science, statistics, and computer science, it's invaluable to be able to deeply listen and be curious and industrious.

My elementary school motto was "deeds not words," and I think this has stuck with me to this day. I've always learnined by doing things; I find talk to be cheap most of the time -- with some notable exceptions for folks such as Ray Kurzweil, Guy Kawazaki, Hans Rosling (R.I.P.), and of course thoughtful, spoken-word maestros such as Bob Dylan. You have to be willing to take yourself out of your comfort zone, to climb the mountain. Jump into the challenge and work around the clock to get the project done.

I think the most important things I learned from my Ph.D. work were how to think and how to climb a mountain, rather than any specific set of techniques.

What's a typical day like for you? Do you work mostly with a team or mostly alone? Which do you prefer?

A good day starts early with a 30-minute walk followed by calls with Europe before their day ends. I juggle a lot of activities in the middle of the day and get more time to think and work solo in the afternoons as people on the East Coast end their days.

I try to maintain a blend of customer and team work, categorized into strategic (months, years), tactical (weeks, months), and operational (right now) tasks. I like to have consistent themes across all the time scales -- a foundation underneath our main workstreams -- when prioritizing my team, customer, and personal schedule. In the end it's a team effort and we all work together -- from field tech to product, engineering, sales, and marketing -- to get the job done. My data science team does the heavy lifting!

Where is data analytics/data science headed in the next few years?

AI and machine learning are the hot area right now and are becoming prevalent in a variety of industries.

I think of AI in three broad groups:

• Smart AI-enabled machines and smart AI-enabled software systems that augment users' intelligence as software and machine operators


• Smart data discovery and visualization, data science, machine learning, modeling, and simulation capabilities at your fingertips for solving business problems


• Embedded models and closed-loop systems of insight that update and drive actions across the business

I believe closed-loop systems of insight are leading the way in digital business transformation as we evolve from smart insight discovery into insight execution, embedded results, and self-learning.

Embedded AI systems are evolving rapidly. I've recently been working on natural language interfaces to analytics systems -- these allow users to speak to devices and obtain visual analyses along with generated narratives addressing a business question. The voice services and APIs available are getting better, and we are setting up abstraction layers and libraries for rich functionality and reuse. Maybe more important is that the general public now has realistic expectations for such interfaces based on interactions with virtual assistants such as Siri, Alexa, and Amy.

I've been around AI and machine learning all of my (now long) career. Previously, public perception of AI felt like science fiction, while my work setting up a simple knowledge-based system was laborious. In this latest wave, the end-user expectations are real, and the software and analytics environments are smart and snappy.