When Cristiana Lara was growing up in Rio de Janeiro, she became obsessed with challenging math problems, working on them feverishly until she could solve them. Now she sees a direct link between a love for puzzling inspired by her technically oriented parents — her father was an electrical engineer and her mother a STEM teacher — and her current work as an Amazon research scientist.
Lara joined Amazon in 2019, after completing her doctorate in process systems engineering at Carnegie Mellon University’s Center for Advanced Process Decision Making. That was the same year Amazon unveiled free, one-day delivery for select Amazon Prime customers in the US. Thus her start date meant Lara began work on a most daunting mission: the development and implementation of an optimization framework to support the company’s transportation network.
While the company has focused on making its delivery systems as efficient and cost-effective as possible, Amazon’s growth, scale, and drive to meet customer demand put a considerable strain on its delivery network. Optimization is a constantly moving target that requires long-term strategic planning, and that is where Lara centers her attention. The focus of her research: develop models and algorithms for solving large-scale discrete optimization problems.
My models and tools actually get to change business decisions and have a direct, positive impact for our customers.
Lara has already made an impact in her short time at Amazon. Her groundbreaking work on network timing, through a planning tool appropriately named ‘TICTOC,’ is helping Amazon to better formulate how to more efficiently transport packages through the “middle mile” of its complex delivery network.
“I particularly like the fact that at Amazon, the work that I do is core to the business,” she said. “My success is not measured by how many papers I publish, but it’s about how my models and tools actually get to change business decisions and have a direct, positive impact for our customers.”
TICTOC is an acronym for Transportation Intraday Capacity planning for Timing Optimization Computation. Lara developed key advancements within it to support timing-related decisions in the transportation network design space.
Used for long-term planning, TICTOC provides the ability to perform sensitivity analysis and understand how different variables in the network design impact the overall delivery speed. With its international network of fulfillment centers, sort centers, and delivery stations, Amazon has built a complex, real-time delivery organization that relies heavily on coordinated timing and an ability to make in-the-moment adjustments in order to fulfill its ambitious customer-delivery promises, and to meet the company’s goal of having 50 percent of all shipments net zero carbon by 2030.
The goal is to understand the tradeoff between transportation costs and delivery speed, and then make more informed decisions regarding the big picture questions that the company will face in the near and distant future. Determining optimal package flow over time in order to maximize one-day delivery while minimizing cost requires a dizzying array of algorithms. When do you schedule the trucks to depart? How many are needed? And when and where are they needed at any given time?
“Those problems are hard to solve because of their discrete nature,” Lara said, “and there’s a lot of theory behind it. They get a lot harder when the problem gets bigger because of the combinatorial explosion.” Solving these problems on a smaller, regional basis is already feasible. But deciphering them for large, continental geographies, such as the entire US, is where the task gets tougher.
For Lara, translating these issues into action items that can have a dramatic impact on the company’s success is “something that I like a lot,” she said. The harder the problem, the more jazzed she is to address it.
After graduating from the Federal University of Rio de Janeiro with a degree in chemical engineering, Lara realized quickly that it was not the discipline that fit her long-range ambitions. Her advisor in Brazil suggested she look into process systems engineering, replete with its modeling and optimization skillset, and she was hooked. As she began her PhD program at Carnegie Mellon, she moved farther away from chemical engineering, and steered more in the direction of operations research. There, the interface between applied math and coding offered a chance to see her work impact decision making in operations, manufacturing, logistics, and a variety of business applications.
Having eschewed offers from academia for a tenure track position, Lara transformed an internship at Amazon into a full-time research job.
“I interned in the same team I currently work for, and my project was to develop a tool combining stochastic simulation and machine learning to forecast the package flow between origin-destination pairs in the Amazon network,” Lara recalled. “The network design optimization models need to consume these kinds of forecasts to be able to plan for the connectivity — how to connect the nodes in the network and how much package flow to expect between nodes and within each node.”
Lara said internships are also a good way for students to figure out their path forward. “My advice is this: Students should take advantage of the opportunity and do as many and as diverse internships as they can. It’s a great way to get to know themselves, what motivates them, the type of working culture that matches their personality, and what they want for their career.”
Having gone that route herself, today Lara finds herself doing work that has real-world impact. To that end, she was nominated by her bosses for an invitation to the prestigious U.S. Frontiers of Engineering symposium, sponsored by the National Academy of Engineering, which was held in Sept. 22 - 24 in Irvine, Calif. Lara presented a poster at the symposium which brought together 83 of the nation’s outstanding young engineers from industry, academia, and government in a variety of disciplines to discuss pioneering technical issues and leading-edge research in various engineering fields.
Dr. Gregory D. Abowd, the dean of Northeastern University’s College of Engineering, was a 2002 participant in the conference. “The purpose is to seed conversations on important global and national problems with a number of smart and open-minded individuals,” Abowd said. “You can say you want to have an impact in the world, but to do so, you have to step out of your discipline and be comfortable thinking on a larger scale.”
The symposium put him in the same room as a group of future leaders in their fields which left him feeling “empowered and emboldened.” For Lara, “It certainly is a vote of confidence from her employer that she has the right kind of expertise and broad-minded, potential leadership capabilities that are worth nurturing,” he added.
“For me, it’s a great opportunity to be among other early career engineers in different fields and be able to talk about my research and their research and learn from them,” Lara said. “Amazon has a lot of researchers and they know that to keep researchers happy, we need to be able to talk about our research, because that’s what excites us.”
