The MIT Machine Intelligence Community emerged from informal gatherings where friends shared pizza while discussing groundbreaking machine learning research. Fast forward three years, and this undergraduate organization has flourished into a vibrant hub of 500 members, featuring an active Slack channel and an impressive array of student-led reading groups and workshops dedicated to demystifying machine learning and artificial intelligence (AI). This academic year marked a significant milestone as MIC partnered with the MIT Quest for Intelligence, combining efforts toward their shared mission of making artificial intelligence model training accessible to all students.
Beginning last autumn, the MIT Quest opened its doors to MIC members, extending valuable access to cloud computing resources donated by industry giants IBM and Google. This partnership dramatically enhanced students' capabilities, elevating them beyond the constraints of running AI models on desktop computers equipped with additional graphics processors. Currently, the MIT Quest and MIC are jointly advancing multiple projects through independent initiatives and collaborations facilitated by MIT's prestigious Undergraduate Research Opportunities Program (UROP).
"We discovered their mission to disseminate machine learning knowledge among all undergraduates and immediately recognized the alignment with our objectives," explains Joshua Joseph, chief software engineer with the MIT Quest Bridge. "We realized we were working toward the same goals and decided to join forces!"
An Innovation Hub for AI Enthusiasts
U.S. Army ROTC students Ian Miller and Rishi Shah initially joined MIC attracted by the complimentary cloud computing credits, but remained engaged after discovering a workshop on neural computing sticks. These compact devices enable mobile devices to perform real-time image processing, inspiring the cadets to realize their vision for a portable computer vision system.
"Without this technology, we would need to transmit images to a centralized location for processing, creating significant logistical challenges," notes Miller, a rising junior. "The neural computing stick eliminated this potential bottleneck."
In just two months and with a modest budget of $200, the team developed a wallet-sized device designed to connect to a tablet mounted on an Army soldier's chest, capable of scanning the environment to identify vehicles and individuals. The students believe that with additional training data, the system could learn to recognize cellphones and weapons. In May, the cadets demonstrated their invention at MIT's Soldier Design Competition and received an invitation from an Army sergeant to continue development at Fort Devens.
Rose Wang, a rising senior specializing in computer science, was also attracted to MIC by the complimentary cloud computing resources and the opportunity to collaborate on projects with the Quest for Intelligence and fellow students. This spring, she utilized IBM cloud credits to execute a reinforcement learning model as part of her research with MIT Professor Jonathan How, focusing on training robot agents to cooperate on tasks involving limited communication and information. She recently presented her findings at a workshop during the prestigious International Conference on Machine Learning.
"The cloud resources enabled me to experiment with various techniques without concerns about computational limitations or resource depletion," she explains. "This freedom significantly accelerated my research progress."
Enhancing AI Accessibility Across Campus
The MIC has initiated several innovative AI projects. Among the most ambitious is Monkey, a container-based, cloud-native platform designed to enable MIT undergraduates to access artificial intelligence model training from any location. The system would track training progress in real-time and manage computing credits allocated to each student. On a Friday afternoon in April, the team convened in a Quest for Intelligence conference room as Michael Silver, a rising senior, outlined the architectural components required for Monkey.
As Silver wrote "Docker Image Build Service" on the whiteboard, the student assigned to research this module expressed regret. "I didn't make significant progress because I was managing three midterm examinations!" he admitted.
The planning session continued with Steven Shriver, a software engineer with the Quest Bridge, contributing valuable insights. The students had assumed Docker, the container service they planned to implement, would provide adequate security. They learned this was not necessarily the case.
"Well, it appears we've identified another critical task," Silver remarked, adding "security" to the growing list on the whiteboard.
Subsequently, the preliminary design would be transformed into a comprehensive document and shared with the two UROP students assisting with Monkey's development. The team aims to launch the platform sometime next year.
"The coding itself isn't the most challenging aspect," explains UROP student Amanda Li, a member of MIC Dev-Ops. "Rather, it's navigating the server-side components of machine learning—including Docker, Google Cloud, and API integration. The most valuable skill I've acquired is how to efficiently design and pipeline a project of this magnitude."
Silver discovered his passion for AI engineering in 2016, when the computer program AlphaGo defeated the world's reigning Go champion. During his senior year at Boston University Academy, Silver explored natural language processing in the laboratory of MIT Professor Boris Katz and has continued collaborating with Katz since enrolling at MIT. Seeking additional coding experience, he transitioned from his role as co-director at HackMIT to join MIC Dev-Ops.
"Many students read about machine learning models but lack practical experience in training them," Silver observes. "Even with theoretical knowledge, the cost of GPUs required for training would amount to several thousand dollars. MIC enables interested students to advance beyond these barriers and gain hands-on experience."
Another student-led initiative focuses on enhancing the exploration of AI research papers available on arXiv. With nearly 14,000 academic papers uploaded monthly, navigating specific subtopics within this vast repository can be overwhelming despite field-specific tagging.
Wang experienced this frustration firsthand while conducting basic literature research on reinforcement learning. "You're confronted with enormous amounts of data without effective methods to organize and present it meaningfully," she notes. "A visualization tool that contextualizes papers and enables exploration based on citation metrics or thematic relevance would have been incredibly valuable."
A third MIC project involves systematically scanning MIT's numerous listservs for AI-related talks and events to populate a dedicated Google calendar. This tool draws inspiration from an application Silver helped develop during MIT's Independent Activities Period in January. Named Dormsp.am, the app categorizes listserv emails sent to MIT undergraduates and organizes them into a calendar-email interface. Students can then search for events by date or color-coded topics such as technology, food, or career opportunities. Following Dormsp.am's launch, Silver will adapt the framework to identify and post AI-related events at MIT to an MIC calendar.
Silver emphasizes that the team devoted considerable attention to user interface design, applying principles from MIT Professor Daniel Jackson's Software Studio course. "An application of this nature succeeds or fails based on its usability, making the front-end development particularly crucial," he explains.
Wang is now collaborating with Moin Nadeem, MIC's outgoing president, to develop the visualization tool. This hands-on experience embodies precisely the practical learning opportunities MIC was created to provide, according to Nadeem, a rising senior. "Students often master theoretical concepts in class without understanding their implementation," he reflects. "I'm working to build the resource I wish had existed during my freshman year: a community of enthusiastic individuals passionate about exploring machine learning through innovative projects."
