What do you do for a next act after launching a new generation of supercomputers that promises to change the game in high-performance computing? If you're IBM's William R. Pulleyblank, you try something very different: Become a consultant. Pulleyblank late last year took over a new project at IBM, the Center for Business Optimization (CBO) in IBM's Business Consulting Services (BCS) unit.
His task is to help businesses and governments use math to solve their most complex problems. In most of these cases, the amount of data they're handling is massive, and supercomputers are required. "My mission is like the Star Trek theme, to go boldly into new areas," says Pulleyblank, 57, who spent most of his 20 years at IBM (IBM ) in research.
Pulleyblank's project is a key piece of IBM's overall business-services strategy. Big Blue figures a good way to avoid being caught in the low-margin commodity-tech business is to harness advanced technology to help organizations fundamentally transform the way they operate.
"Many clients have done the low-hanging fruit in terms of performance optimization. But to get the competitive advantage, they need to get to the next level," says Ginni Rometty, managing partner of BCS and Pulleyblank's boss.
IBM has for years been assigning mathematicians to help clients with complex problems, but Pulleyblank's CBO group makes it a formal business initiative. He has already rounded up 25 "black-belt" IBM consultants with deep knowledge of individual industries and their operational challenges. And he plans to hire an additional 15 to 25. His group taps into a few hundred scientists at IBM Research who have math and data-analysis expertise. The BCS sales force takes their offerings to clients.
An engagement with the U.S. Postal Service -- one of the group's first -- shows the power of math and what it can do. The Postal Service contracted with PricewaterhouseCoopers Consulting more than three years ago to assist with modeling its mail-processing and transportation networks. The consultants brought IBM researchers into the project after IBM bought PWCC two years ago.
Now, Pulleyblank's crew is on the case. The challenge: designing a flexible logistics network that reduces total costs, makes operations more efficient, and improves deliveries so the Postal Service can compete with UPS (UPS ), Federal Express (FDX ), and other commercial mail shippers. "This is a key part of the postal transformation plan," says Pranab Shah, the Postal Service's manager of network-operations development.
First, a new state-of-the-art mail-processing system was created. Finished last year, it awaits approval by Postal Service management. The next phase is perfecting the shipping of mail between the Postal Service's 300 distribution centers.
IBM researchers late last year completed a simulation of the entire system. They're now working on a program postal managers can use day-to-day to decide how best to move pieces of mail -- whether by truck, commercial airline, or dedicated flights.
Pulleyblank aims to tap the work his group does for specific clients and turn it into consulting templates, software, or even managed services. He doesn't want to have to assign his black belts to every new client and start from scratch. "The challenge will be taking the knowledge and reusing it for other clients -- but avoid getting too general with the first client," says Pulleyblank.
He's used to thorny challenges that take a long time to solve. IBM's Blue Gene supercomputer, which he directed, took five years to go from concept to delivery late last year. Life as a consultant should be a cinch compared to that.
By Steve Hamm in New York