Building Better in the Cloud: Why the Time Is Now

Massive cloud investment continues worldwide, with Gartner predicting public cloud spending to reach an eye-popping $1 trillion by 2027. This number is growing significantly as companies invest more in generative AI, as GenAI initiatives require a lot of cloud capacity. And yet, many organizations still struggle…

New Wave Technology Makes Android Emotions More Natural

For those who have interacted with an android that looks incredibly human, many report that something “feels off.” This phenomenon goes beyond mere appearance – it is deeply rooted in how robots express emotions and maintain consistent emotional states. Or in other words, their lack of…

Daniel Cane, Co-CEO and Co-Founder of ModMed – Interview Series

Daniel Cane is co-CEO and cofounder of South Florida-based ModMed®, a healthcare IT company that is transforming healthcare through specialty-specific, intelligent platforms to increase practice efficiency and improve patient outcomes. Founded in February 2010, ModMed has grown to over 1,200 employees and has raised over $332…

High school teams compete at 2024 MIT Science Bowl Invitational

A quiet intensity held the room on edge as the clock ticked down in the final moments of the 2024 MIT Science Bowl Invitational. Montgomery Blair High School clung to a razor-thin lead over Mission San Jose High School — 70 to 60 — with just two minutes remaining.

Mission San Jose faced a pivotal bonus opportunity that could tie the score. The moderator’s steady voice filled the room as he read the question. Mission San Jose’s team of four huddled together, pencils moving quickly across their white scratch paper. Across the stage, Montgomery Blair’s players sat still, their eyes darting between the scoreboard and the opposing team attempting to close the gap.

Mission San Jose team captain Advaith Mopuri called out their final answer.

“Incorrect,” the moderator announced.

Montgomery Blair’s team collectively exhaled, the tension breaking as they sealed their championship victory, but the gravity of those final moments when everything was on the line lingered — a testament to just how close the competition had been. Their showdown in the final round was a fitting culmination of the event, showcasing the mental agility and teamwork honed through months of practice.

“That final round was so tense. It came down to the final question,” says Jonathan Huang, a senior undergraduate at MIT and the co-president of the MIT Science Bowl Club. “It’s rare for it to come down to the very last question, so that was really exciting.”​

A tournament of science and strategy

Now in its sixth year at the high school level, the MIT Science Bowl Invitational welcomed 48 teams from across the country this year for a full day of competition. The buzzer-style tournament challenged students on topics that spanned disciplines such as biology, chemistry, and physics. The rapid pace and diverse subject matter demanded a combination of deep knowledge, quick reflexes, and strategic teamwork.

Montgomery Blair’s hard-fought victory marked the culmination of months of preparation. “It was so exciting,” says Katherine Wang, Montgomery Blair senior and Science Bowl team member. “I can’t even describe it. You never think anything like that would happen to you.”

The volunteers who make it happen

Behind the scenes, the invitational is powered by a team of more than 120 dedicated volunteers, many of them current MIT students. From moderating matches to coordinating logistics, these volunteers form the backbone of the invitational.

Preparation for the competition starts months in advance. “By the time summer started, we already had to figure out who was going to be the head writers for each subject,” Huang says. “Every week over the summer, volunteers spent their own time to start writing up questions.”

“Every single question you hear today was written by a volunteer,” said Paolo Adajar, an MIT graduate student who served in roles like questions judge this year and is a former president of the MIT Science Bowl Club. Adajar, who competed in the National Science Bowl as a high school student, has been involved in the MIT Invitational since it began in 2019. “There’s just something so fun about the games and just watching people be excited to get a question right.”

For many volunteers, the event is a chance to reconnect with a shared community. “It’s so nice to get together with the community every year,” says Emily Liu, a master’s student in computer science at MIT and a veteran volunteer. “And I’m always pleasantly surprised to see how much I remember.”

Looking ahead

For competitors, the invitational offers more than just a chance to win. It’s an opportunity to connect with peers who share their passion for science, to experience the energy of MIT’s campus, and to sharpen skills they’ll carry into future endeavors. 

As the crowd dispersed and the auditorium emptied, the spirit of the competition remained — a testament to the dedication, curiosity, and camaraderie that define the MIT Science Bowl Invitational.

Remembering Mike Walter: “We loved him, and he loved us”

Michael “Mike” Walter, MIT Health applications support generalist, passed away on Nov. 2 at age 46 after a battle with cancer. 

At home, Walter was a husband and devoted father to his two adolescent sons. But for 22 years, he was everyone’s friend and the smiling face at MIT Health who never failed to solve individual computer problems, no matter how large or small. 

Walter came to MIT as an office assistant in MIT Health’s Medical Records department in 2002. He eventually transferred to MIT Health’s Technology Services team, where he worked from 2009 until his passing. Information Systems Manager David Forristall, who had previously worked in medical records, still remembers when “this young guy came to work for his first day.”  

“When he first got to Medical Records, he thought it was only going to be a pit stop — that he was only going to be here for like two weeks,” says Walter’s colleague, Technical Support Specialist Michael Miller. “Then, 20 years later…” 

“You don’t often, other than a family member, watch someone grow through their life,” says Forristall. “So for him to come to MIT as a young man at the start of his career, to a full-blown career with a wife and children. He basically came here as a boy, and we watched him turn into a man.” 

Walter’s colleagues were always struck by how positive he was. “He never complained about help desk tickets. All of us looked to him for that,” remembers Medical Records Manager Tom Goodwin. “When I found myself getting a little annoyed, I would just look to Mike and think, he doesn’t do that.” 

Without fail, Walter would drop everything to help his MIT Health colleagues. “He would go out on a call, and people would just keep stopping him,” remembers Senior Programmer Analyst Terry McNatt. “They would see him around the building, and they knew he would help them. He wouldn’t come back for two hours!”

The needs of MIT patients were just as important to Walter. At the annual flu clinics, Walter would, without fail, volunteer for the full day. Oftentimes people could find him serving as a go-fer; he would deliver vaccines, Band-Aids, and whatever other supplies were needed to help the vaccinators be as efficient as possible.  

According to his colleagues, Walter’s dedication to the MIT community is best explained by the day he learned of his cancer diagnosis. A major snowstorm was approaching, and Walter was diligently working to get laptop computers set up so employees could work remotely for multiple days if needed. All the while, he felt awful. Eventually he went to Urgent Care to be seen.  

“Urgent Care was telling him, ‘You need to go to Mount Auburn hospital right now,’” recalls Forristall. “But Mike didn’t want to go.” He refused to leave until all the laptops were properly set up so his colleagues could continue to care for patients despite the impending MIT snow closure. He only left after he grudgingly agreed to have his peers cover for him.  

Walter was also a Patriots superfan, and deep lover of sports. He had multiple footballs at his desk at all times, and for years he would gather his colleagues for “coffee-break” walks around campus where they would all walk and toss a football back and forth. Anyone who passed by was invited to Walter’s game of catch — students, construction workers, staff, and faculty alike were welcome.  

“Mike was always happy and he shared that with everyone,” says Forristall. “He made you happy when you saw him. We loved him and he loved us.”  

Mike Walter is survived by his wife Cindy (Cucinotta), his sons Ben and Leo, and many extended family members and friends. See his legacy page here.

A new computational model can predict antibody structures more accurately

By adapting artificial intelligence models known as large language models, researchers have made great progress in their ability to predict a protein’s structure from its sequence. However, this approach hasn’t been as successful for antibodies, in part because of the hypervariability seen in this type of protein.

To overcome that limitation, MIT researchers have developed a computational technique that allows large language models to predict antibody structures more accurately. Their work could enable researchers to sift through millions of possible antibodies to identify those that could be used to treat SARS-CoV-2 and other infectious diseases.

“Our method allows us to scale, whereas others do not, to the point where we can actually find a few needles in the haystack,” says Bonnie Berger, the Simons Professor of Mathematics, the head of the Computation and Biology group in MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL), and one of the senior authors of the new study. “If we could help to stop drug companies from going into clinical trials with the wrong thing, it would really save a lot of money.”

The technique, which focuses on modeling the hypervariable regions of antibodies, also holds potential for analyzing entire antibody repertoires from individual people. This could be useful for studying the immune response of people who are super responders to diseases such as HIV, to help figure out why their antibodies fend off the virus so effectively.

Bryan Bryson, an associate professor of biological engineering at MIT and a member of the Ragon Institute of MGH, MIT, and Harvard, is also a senior author of the paper, which appears this week in the Proceedings of the National Academy of Sciences. Rohit Singh, a former CSAIL research scientist who is now an assistant professor of biostatistics and bioinformatics and cell biology at Duke University, and Chiho Im ’22 are the lead authors of the paper. Researchers from Sanofi and ETH Zurich also contributed to the research.

Modeling hypervariability

Proteins consist of long chains of amino acids, which can fold into an enormous number of possible structures. In recent years, predicting these structures has become much easier to do, using artificial intelligence programs such as AlphaFold. Many of these programs, such as ESMFold and OmegaFold, are based on large language models, which were originally developed to analyze vast amounts of text, allowing them to learn to predict the next word in a sequence. This same approach can work for protein sequences — by learning which protein structures are most likely to be formed from different patterns of amino acids.

However, this technique doesn’t always work on antibodies, especially on a segment of the antibody known as the hypervariable region. Antibodies usually have a Y-shaped structure, and these hypervariable regions are located in the tips of the Y, where they detect and bind to foreign proteins, also known as antigens. The bottom part of the Y provides structural support and helps antibodies to interact with immune cells.

Hypervariable regions vary in length but usually contain fewer than 40 amino acids. It has been estimated that the human immune system can produce up to 1 quintillion different antibodies by changing the sequence of these amino acids, helping to ensure that the body can respond to a huge variety of potential antigens. Those sequences aren’t evolutionarily constrained the same way that other protein sequences are, so it’s difficult for large language models to learn to predict their structures accurately.

“Part of the reason why language models can predict protein structure well is that evolution constrains these sequences in ways in which the model can decipher what those constraints would have meant,” Singh says. “It’s similar to learning the rules of grammar by looking at the context of words in a sentence, allowing you to figure out what it means.”

To model those hypervariable regions, the researchers created two modules that build on existing protein language models. One of these modules was trained on hypervariable sequences from about 3,000 antibody structures found in the Protein Data Bank (PDB), allowing it to learn which sequences tend to generate similar structures. The other module was trained on data that correlates about 3,700 antibody sequences to how strongly they bind three different antigens.

The resulting computational model, known as AbMap, can predict antibody structures and binding strength based on their amino acid sequences. To demonstrate the usefulness of this model, the researchers used it to predict antibody structures that would strongly neutralize the spike protein of the SARS-CoV-2 virus.

The researchers started with a set of antibodies that had been predicted to bind to this target, then generated millions of variants by changing the hypervariable regions. Their model was able to identify antibody structures that would be the most successful, much more accurately than traditional protein-structure models based on large language models.

Then, the researchers took the additional step of clustering the antibodies into groups that had similar structures. They chose antibodies from each of these clusters to test experimentally, working with researchers at Sanofi. Those experiments found that 82 percent of these antibodies had better binding strength than the original antibodies that went into the model.

Identifying a variety of good candidates early in the development process could help drug companies avoid spending a lot of money on testing candidates that end up failing later on, the researchers say.

“They don’t want to put all their eggs in one basket,” Singh says. “They don’t want to say, I’m going to take this one antibody and take it through preclinical trials, and then it turns out to be toxic. They would rather have a set of good possibilities and move all of them through, so that they have some choices if one goes wrong.”

Comparing antibodies

Using this technique, researchers could also try to answer some longstanding questions about why different people respond to infection differently. For example, why do some people develop much more severe forms of Covid, and why do some people who are exposed to HIV never become infected?

Scientists have been trying to answer those questions by performing single-cell RNA sequencing of immune cells from individuals and comparing them — a process known as antibody repertoire analysis. Previous work has shown that antibody repertoires from two different people may overlap as little as 10 percent. However, sequencing doesn’t offer as comprehensive a picture of antibody performance as structural information, because two antibodies that have different sequences may have similar structures and functions.

The new model can help to solve that problem by quickly generating structures for all of the antibodies found in an individual. In this study, the researchers showed that when structure is taken into account, there is much more overlap between individuals than the 10 percent seen in sequence comparisons. They now plan to further investigate how these structures may contribute to the body’s overall immune response against a particular pathogen.

“This is where a language model fits in very beautifully because it has the scalability of sequence-based analysis, but it approaches the accuracy of structure-based analysis,” Singh says.

The research was funded by Sanofi and the Abdul Latif Jameel Clinic for Machine Learning in Health. 

Unlocking the hidden power of boiling — for energy, space, and beyond

Most people take boiling water for granted. For Associate Professor Matteo Bucci, uncovering the physics behind boiling has been a decade-long journey filled with unexpected challenges and new insights.

The seemingly simple phenomenon is extremely hard to study in complex systems like nuclear reactors, and yet it sits at the core of a wide range of important industrial processes. Unlocking its secrets could thus enable advances in efficient energy production, electronics cooling, water desalination, medical diagnostics, and more.

“Boiling is important for applications way beyond nuclear,” says Bucci, who earned tenure at MIT in July. “Boiling is used in 80 percent of the power plants that produce electricity. My research has implications for space propulsion, energy storage, electronics, and the increasingly important task of cooling computers.”

Bucci’s lab has developed new experimental techniques to shed light on a wide range of boiling and heat transfer phenomena that have limited energy projects for decades. Chief among those is a problem caused by bubbles forming so quickly they create a band of vapor across a surface that prevents further heat transfer. In 2023, Bucci and collaborators developed a unifying principle governing the problem, known as the boiling crisis, which could enable more efficient nuclear reactors and prevent catastrophic failures.

For Bucci, each bout of progress brings new possibilities — and new questions to answer.

“What’s the best paper?” Bucci asks. “The best paper is the next one. I think Alfred Hitchcock used to say it doesn’t matter how good your last movie was. If your next one is poor, people won’t remember it. I always tell my students that our next paper should always be better than the last. It’s a continuous journey of improvement.”

From engineering to bubbles

The Italian village where Bucci grew up had a population of about 1,000 during his childhood. He gained mechanical skills by working in his father’s machine shop and by taking apart and reassembling appliances like washing machines and air conditioners to see what was inside. He also gained a passion for cycling, competing in the sport until he attended the University of Pisa for undergraduate and graduate studies.

In college, Bucci was fascinated with matter and the origins of life, but he also liked building things, so when it came time to pick between physics and engineering, he decided nuclear engineering was a good middle ground.

“I have a passion for construction and for understanding how things are made,” Bucci says. “Nuclear engineering was a very unlikely but obvious choice. It was unlikely because in Italy, nuclear was already out of the energy landscape, so there were very few of us. At the same time, there were a combination of intellectual and practical challenges, which is what I like.”

For his PhD, Bucci went to France, where he met his wife, and went on to work at a French national lab. One day his department head asked him to work on a problem in nuclear reactor safety known as transient boiling. To solve it, he wanted to use a method for making measurements pioneered by MIT Professor Jacopo Buongiorno, so he received grant money to become a visiting scientist at MIT in 2013. He’s been studying boiling at MIT ever since.

Today Bucci’s lab is developing new diagnostic techniques to study boiling and heat transfer along with new materials and coatings that could make heat transfer more efficient. The work has given researchers an unprecedented view into the conditions inside a nuclear reactor.

“The diagnostics we’ve developed can collect the equivalent of 20 years of experimental work in a one-day experiment,” Bucci says.

That data, in turn, led Bucci to a remarkably simple model describing the boiling crisis.

“The effectiveness of the boiling process on the surface of nuclear reactor cladding determines the efficiency and the safety of the reactor,” Bucci explains. “It’s like a car that you want to accelerate, but there is an upper limit. For a nuclear reactor, that upper limit is dictated by boiling heat transfer, so we are interested in understanding what that upper limit is and how we can overcome it to enhance the reactor performance.”

Another particularly impactful area of research for Bucci is two-phase immersion cooling, a process wherein hot server parts bring liquid to boil, then the resulting vapor condenses on a heat exchanger above to create a constant, passive cycle of cooling.

“It keeps chips cold with minimal waste of energy, significantly reducing the electricity consumption and carbon dioxide emissions of data centers,” Bucci explains. “Data centers emit as much CO2 as the entire aviation industry. By 2040, they will account for over 10 percent of emissions.”

Supporting students

Bucci says working with students is the most rewarding part of his job. “They have such great passion and competence. It’s motivating to work with people who have the same passion as you.”

“My students have no fear to explore new ideas,” Bucci adds. “They almost never stop in front of an obstacle — sometimes to the point where you have to slow them down and put them back on track.”

In running the Red Lab in the Department of Nuclear Science and Engineering, Bucci tries to give students independence as well as support.

“We’re not educating students, we’re educating future researchers,” Bucci says. “I think the most important part of our work is to not only provide the tools, but also to give the confidence and the self-starting attitude to fix problems. That can be business problems, problems with experiments, problems with your lab mates.”

Some of the more unique experiments Bucci’s students do require them to gather measurements while free falling in an airplane to achieve zero gravity.

“Space research is the big fantasy of all the kids,” says Bucci, who joins students in the experiments about twice a year. “It’s very fun and inspiring research for students. Zero g gives you a new perspective on life.”

Applying AI

Bucci is also excited about incorporating artificial intelligence into his field. In 2023, he was a co-recipient of a multi-university research initiative (MURI) project in thermal science dedicated solely to machine learning. In a nod to the promise AI holds in his field, Bucci also recently founded a journal called AI Thermal Fluids to feature AI-driven research advances.

“Our community doesn’t have a home for people that want to develop machine-learning techniques,” Bucci says. “We wanted to create an avenue for people in computer science and thermal science to work together to make progress. I think we really need to bring computer scientists into our community to speed this process up.”

Bucci also believes AI can be used to process huge reams of data gathered using the new experimental techniques he’s developed as well as to model phenomena researchers can’t yet study.

“It’s possible that AI will give us the opportunity to understand things that cannot be observed, or at least guide us in the dark as we try to find the root causes of many problems,” Bucci says.

Protecting Your AI Investment: Why Cooling Strategy Matters More Than Ever

Data center operators are gambling millions on outdated cooling technology. The conversation around data center cooling isn’t just changing—it’s being completely redefined by the economics of AI. The stakes have never been higher. The rapid advancement of AI has transformed data center economics in ways few…

Anthony Tobelaim, Co-founder & CPO at Siit – Interview Series

Anthony Tobelaim is the Co-founder & CPO of Siit. He previously worked at Aircall as a Senior Director of Partnership EMEA & APAC. Siit empowers IT and internal operations teams with a modern platform designed to simplify and streamline service management processes from start to finish….