Profiles in OR/MS: Harlan Crowder
|
Optimization Consultant Manager,
B.S. Chemistry and Biology, |
PhD Computer Science,
Contact Information: |
Questions & Answers
Q. How did you arrive at your current job with ILOG?
A. I had put in a full career at IBM as a researcher, software developer and consultant. I was looking for an opportunity that allowed me to use my experience in a new setting. I found ILOG especially attractive because of the interesting range of optimization products and the top quality of the people. They were a fast-growing company and I was convinced I could make a contribution to their success.
Q. How large is ILOG and what are the main products and services the company provides?
A. ILOG has 450 people world-wide, with the fastest growth in the US. We sell high-performance software components for optimization, visualization and application architecture. We also have a thriving consulting business for helping companies use our technology. For people interested in optimization technology, our most well know product is CPLEX, the industry standard system for solving linear programming and related problems.
Q. What is your main role as Optimization Consulting Manager at ILOG?
A. I lead a group of about 10 optimization consultants. We use the full range of ILOG optimization products for solving analytic business applications. In addition to CPLEX, we use the ILOG suite of constraint programming tools for attacking difficult combinatorial optimization problems such as scheduling, sequencing and logistics. We have two main classes of clients for our optimization consulting services:
- Independent software vendors – Companies that are producing sophisticated software applications that have an optimization component. We help design and implement the optimization modeling portions of their systems. Examples include I2 Technologies, PeopleSoft, SAP, and Aspen Technologies.
- Industrial optimization users – Individual users of optimization technology. We specialize in sophisticated applications, especially when they stretch the envelope of our products. Our clients include American Airlines, Chrysler, Boeing, Federal Express and Aeroports de Paris.
The types of problems we address spans the range of OR/MS applications, including supply chain planning and scheduling, staff scheduling, vehicle routing, job-shop scheduling and capital budgeting.
Q. Do customers approach ILOG for help? In other words, do you find that more customers are becoming aware of the need for operations research techniques in order to enhance their products and stay competitive?
A. ILOG typically combines consulting services with sales of our technology. Most of our engagements involve designing and implementing a prototype application that solves a specific class of problem. Occasionally we will help clients build production systems that employee our optimization technology. Our “leads” for optimization consulting projects come from a number of sources, primarily from our ILOG sales force, but also from previous client referrals, word of mouth and personal contacts with our consultants. Increasingly, we are getting leads for consulting directly from the Internet.
Q. Prior to your employment at ILOG, what are some of the past OR/MS projects that you found especially memorable and rewarding?
A. Prior to joining ILOG, I was with the Management Technologies Consulting Group at IBM Global Services. We helped a variety of companies solve problems in manufacturing, logistics, finance and marketing. Among my most memorable O.R.-related engagements were:
- Helping a steel company in Shanghai implement a production planning system. This company was emerging from state ownership and most of the people had little knowledge of costs, profits and other basic financial metrics that we take for granted. I ended up teaching them an Economics 101 class.
- Designing and implementing a global supply chain model for an old-line New England maker of hand tools. They where trying to compete in a global market using methods that were appropriate for a regional manufacturer. Using analytic business techniques such as optimization was a real eye-opener for their management.
- Designing and building a sophisticated optimization model for America’s top volume direct mail company. We helped them significantly reduce their advertising budget (and cut down on the number of catalogs they pushed through the mail) while increasing their overall profit margins. The application involved a computationally challenging column generation linear programming approach that buckled the knees of a large multiprocessor supercomputer.
Q. What were the main tools and OR/MS techniques that were used on these projects?
A. In these and many other cases, the tools we used were off-the-shelf modeling and analytic problem solving tools.
Q. What are the most valuable technical skills that you believe are needed to be successful in the OR/MS industry?
A. OR/MS is involved in building models – abstractions of reality – that mimic the operation and characteristics of real-world processes. Modeling involves looking at the process, deciding what is important, and translating the notion into a formal mathematical structure. So if you want to do OR/MS, you need to be able to build good mathematical models of real, often complex systems.
The challenge in modeling is to make sure you are solving the right problem, at the right level of detail. This requires combining the skills of a detective, doctor, mathematician, engineer and, occasionally, a psychiatrist. You are trying to avoid building an elegant and efficient model of the wrong problem. Even after you identify the problem, the critical element is deciding what to model and, more importantly, what to leave out. Too much detail and the exercise becomes hopelessly entangled in complexity; not enough detail and the model solutions are useless for decision making.
The good news is that model building is a skill that can be learned. And the best way to learn is to start building models. Nobody ever learned to ride a bicycle by reading about riding a bicycle. Model building is a skill that needs to be learned and practiced.
For more information about helping people solve the right problem and other issues related to applying OR/MS in practice, see my paper “Seven Helpful Hints for OR/MS Consultants” that appeared in the February 1997 issue of OR/MS Today.
Q. What are some of the most valuable non-technical skills that you believe are needed to be successful in the OR/MS industry?
A. The two most valuable skills for this business are teamwork and communication.
Most important projects involve a team working together to accomplish a set of goals. OR/MS applications don’t occur in a vacuum; they are usually one component of a multidisciplinary effort. A typical application of OR/MS in an industrial setting will involve a team with several skills: business expertise in the particular application, finance, engineering, information systems, and OR/MS. The ability to work in a team environment is critical element in the success of such efforts.
Most clients of OR/MS applications don’t have a good understanding of our discipline, so it is absolutely critical that OR/MS practitioners have good communication skills. Every aspect of the OR/MS application process—gathering requirements, designing models, validating results, deploying applications—involve the communication of facts and ideas to and from other people (see above – solving the right problem). With enough training and practice, anyone can sit in a room and construct elegant OR/MS models. The challenge is to open an effective two-way communication link to a potential user who has a problem.
Q. In what ways do you continue to expand your knowledge of new technologies and techniques in OR/MS?
A. I try to read about new developments and attend seminars at local universities and professional organizations. But mostly I try to hang around with people that are inventing and using new technology.
For example, there is a big push right now at ILOG to better understand how linear programming and constraint programming can effectively work together. The idea is to apply cooperative approaches using both technologies to yield better results than using each technology individually. Our early results have been intriguing for a class of difficult problems.
For example, one of our consultants recently implemented an airline crew scheduling application using this cooperative approach. The “pairing generation” step was implemented using constraint programming. This component generates a large collection of legal crew-route pairings, the rules for which are expressed in a constraint programming representation. The results of crew pairing are then inserted into a large integer programming problem for selecting the best crew schedules. The cooperative approach allows both components of the crew scheduling application to designed and implemented as an integrated system.
For more information on ILOG optimization technology and applications, see the optimization white paper.
Q. What do you find most rewarding about your career in OR/MS?
A. For me the whole purpose of OR/MS is to help people build sophisticated tools for making good decisions. I get a good feeling when I’m helping people achieve that goal. A couple of years ago I spent some time in Taiwan helping a fast-growing high-tech manufacturer understand and model their production process. They really appreciated that I had come half way around the world to help them with a complicated task. I contributed to their success and that was a rewarding experience.
Q. What advice do you have for those just starting out in a career in OR/MS?
A. If you look at who is successful in our discipline, you usually find a common thread: they have learned how to do one thing and they have learned how to do it very well. There are a lot of Technological Temptations out there and it is easy to get distracted by trying to gain knowledge and experience in too many things. It’s OK to broaden your scope, but first become an expert in something.
Q. What do you predict the future has in store for the field of OR/MS and for OR/MS practitioners?
A. Most traditional real-world OR/MS applications involve improving the bottom line of an enterprise by attacking and cutting the cost structure of critical processes. Recently, there has been a significant increase in applications aimed at improving the bottom line by increasing revenue and better use of scarce resources. Of course when computing the bottom line, cutting a dollar of cost is equal to gaining a dollar in revenue. But managers and executives give a much better reception to someone who wants to help make them a million dollars (“Tell me more!”) than to someone who claims to be able to cut a million in costs (“Hmmm, maybe, maybe not”).
I predict that in the face of increased complexity and uncertainty in global commercial enterprises, we will see an upsurge in OR/MS applications that address revenue and yield management and the allocation of valuable resources for securing a competitive advantage. We have only scratched to surface in understanding how to design, build and deploy these applications.