“I didn’t imagine I could grow and learn so much”
Donato Crisostomi talks about how his mother helped spark a love of knowledge that led him to two science internships at Amazon.
From an early age, Donato Crisostomi was fascinated with mathematics and philosophy. That interest was nurtured by his mother — and eventually helped him land two science internships with Amazon.
As a child growing up in Civitavecchia, a small coastal city near Rome, he enjoyed exploring his natural environment.
“I spent all my time running around in nature poking at leaves, animals, all sorts of things. Each of these encounters filled my mind with questions,” he recalled.
A curious young boy, he looked to his mother for answers.
“She always did her best to answer my questions,” he remembered. While his mother did not have the opportunity to pursue higher education, Crisostomi felt she would have made a great engineer or scientist if given the chance.
“She understood, and probably still does, how an electronic or mechanical device works better than anyone in the house, including me,” he said.
His mother’s enthusiasm helped spark a passion for learning that has served him well.
Forging his own path
As a first-generation college student, Crisostomi forged his own path toward higher education. Yet it took time for Crisostomi to discover his own abilities. After dropping out of high school and spending a year on a working holiday in Australia, he returned to Italy. He graduated high school, spent a year in the Italian army, and gained the clarity he needed about his long-term goals: He wanted to study computer science.
Computer science provided Crisostomi a platform to indulge his passion for knowledge.
“Computer science has deep philosophical roots,” he stated. “But I was most interested in the ones regarding artificial intelligence.”
Crisostomi went on to earn bachelor’s and master’s degrees in computer science from Sapienza Università di Roma. While finishing his master’s in May 2021, Crisostomi joined the Alexa Natural Language Understanding organization as an intern at Amazon’s Turin office.
He had been following Amazon on LinkedIn since his first year in college as his goal was to conduct applied research.
“I was trying to choose between the dynamic environment of a new start-up or a larger, more consolidated company,” he said. “From the information I gathered online, Amazon’s ‘Day One’ philosophy was a perfect compromise.”
The thought of working at Amazon appeared daunting at first, but Crisostomi’s experience as an intern eased his anxiety.
“I was afraid that I would find a cold environment as it was my first experience in such a huge corporation,” he recalled. “I had some expectations coming from movies I had seen that I would have to talk in front of hundreds of people who would be staring at me and judging me.”
Instead, he found a welcoming and encouraging environment, particularly when it came to his manager Davide Bernardi, a data science manager at Alexa AI Natural Understanding.
“It was really lovely working with Davide and the team at Amazon,” Crisostomi said. “Everyone was so friendly; Davide was so nice, and encouraged me to share my ideas.”
As part of Bernardi’s team, Crisostomi worked with research scientists and developers to automate the creation of artificial cross-lingual datasets to test a neural model against complex multilingual phenomena.
“We wanted to design a neural model which could recognize code mixed utterances,” Crisostomi explained. Code mixing refers to the use of multiple languages in the same utterance – common among bilingual and multilingual speakers. The challenge: have the neural model understand different languages and dialects in the same utterances.
Crisostomi had an idea to make the neural model’s multilingual capacity more efficient by reducing repeated computation during training and removing duplication at the batch level while keeping the data distribution perceived by the model untouched.
Bernardi said he noticed how quickly Crisostomi became proficient in the tools, platforms, techniques, and models.
“We had him work on a big goal when we were already in the middle of the path,” he recalled, “But Donato wasn’t scared. He saw the opportunity to contribute to something big and did it very well.”
Going back to Amazon
Crisostomi, who is pursuing his doctorate in computer science at Sapienza, was so enthralled with his first internship experience, he wrote a LinkedIn post about it back in December.
Difference is appreciated at Amazon. If they have chosen you for the position, then they want you to be who you are, not a copy of others.
“Words will probably fail me to express my gratitude for the time spent here at Amazon, as I didn't imagine I could grow and learn so much in such a brief time,” he wrote. “I would like to dedicate a special thank you from the bottom of my heart to my team in Turin, who made me feel at home while I was far from mine. I am really going to miss you, and hope that our paths will cross again in the future.”
That wish, as it turns out, became reality. Crisostomi recently accepted another six-month internship at TEN Search, Amazon’s Luxembourg location.
There, he is focusing on research with an interdisciplinary information retrieval science team engaged in machine learning, natural language processing, transfer learning, ranking, and software development, with the aim of improving the customer experience.
“I feel the same excitement that I felt as a freshman,” Crisostomi remarked. “Providing a pleasant information retrieval experience to the user is as important as it is hard because it requires overcoming language ambiguities. Having the opportunity to assist thrills me.”
That drive to assist extends to offering his advice to prospective interns. “Totally go for it and don’t be afraid of asking questions,” he said. “If you have a chance to visit the office, go and meet people. And this may sound like a cliché, but be yourself. Difference is appreciated at Amazon. If they have chosen you for the position, then they want you to be who you are, not a copy of others.”
It’s a lesson his mother would certainly approve of.