Most people do not spend much time thinking about the electrical grid. It is only when the lights go out that we remember how much of modern life depends on it.<\/p>\n
Artificial Intelligence works much the same way. The public sees the chatbots, copilots, and AI agents. What they rarely see are the systems underneath them: the cloud platforms, data pipelines, databases, and distributed infrastructure that make those applications possible. But who should build it, and how?<\/p>\n
That question sits at the heart of a new program Boston University has built around AI. Dr. Evimaria Terzi is a Professor of Computer Science at Boston University and Associate Chair for Academics in the Computer Science Department, with a faculty appointment in BU\u2019s Faculty of Computing & Data Sciences. She is one of the architects of the new Online Master of Science in Computer Science & AI<\/a>, and she frames the program by what sits beneath it. \u201cThe focus of the computer science online master\u2019s is very much in the infrastructure part,\u201d she said in a recent interview. \u201cHow to build systems to handle all this big data so that someone can ask queries, or retrieve the data effectively, or run the models in production.\u201d<\/p>\n The infrastructure layer of AI, when it works, is invisible. When it fails, everything stops.<\/p>\n At 11:49 PM Pacific on October 19, 2025, Amazon Web Services began a cascading failure<\/a> in its US-EAST-1 region that took roughly 15 hours to fully resolve. The root cause was not glamorous: a DNS race condition inside DynamoDB\u2019s internal management system. Snapchat, Roblox, Ring, and other services experienced disruptions. The failure was not caused by AI itself, but by the infrastructure underneath it: cloud systems, databases, networking, and orchestration layers. When that infrastructure fails, everything built on top of it fails too.<\/p>\n That is the layer the Online MS in Computer Science & AI is built around.<\/p>\n Anyone reading the labor market in 2026 has seen the same headlines. Junior developer jobs are disappearing, and questions like \u201cwill AI replace software engineers?\u201d are regularly debated. According to Stanford University\u2019s 2026 AI Index Report<\/a>, software developer employment among workers aged 22 to 25 has fallen by approximately 20% since 2024. AI is doing the work that entry-level engineers used to do: routine code generation, standard CRUD operations, scripted testing, basic configuration. All of it is becoming a low-cost machine task.<\/p>\n But the same report shows the opposite trend at a different layer of the stack. Demand for AI engineers, machine learning engineers, AI infrastructure engineers, MLOps engineers, and cloud architects continues to outrun supply. AI-related skills are now requested in 2.5% of all U.S. job postings, a 297% increase over the past decade. Mentions of Agentic AI in job descriptions grew more than 280% in a single year.\u00a0As AI automates more routine technical work, demand appears to be shifting toward engineers with deeper expertise in systems, infrastructure, and architecture.<\/p>\n Terzi has heard this firsthand. Earlier this year, BU\u2019s Computer Science faculty hosted a panel of senior engineers from Amazon, Google, and other major employers. One sentence stayed with her. \u201cThe more things AI does for you as an engineer,\u201d she paraphrased, \u201cthe more knowledgeable you need to be about how the underlying systems work.\u201d<\/p>\n That sentence has provided inspiration for the program.<\/p>\n For many prospective students, \u201cComputer Science\u201d still evokes algorithms, programming languages, and theoretical foundations. Those topics remain important, but modern AI systems also depend on cloud infrastructure, distributed computing, databases, networking, security, and large-scale systems engineering. The Online MS in Computer Science & AI emphasizes that operational layer of computing and was built around a deliberate shift in BU\u2019s department, a recent wave of hiring focused on what Terzi calls \u201csystems people,\u201d faculty who design and build the actual systems that run in the world rather than reasoning about them in the abstract.<\/p>\n \u201cDoing theory of Computer Science online doesn\u2019t really work,\u201d she said. \u201cBut doing systems work online, that actually works very well. And the need in industry is mostly about computer systems and how they integrate AI.\u201d<\/p>\n The curriculum reflects that bet. Students learn how to design cloud platforms, how large-scale data is replicated and retrieved, how to architect for GPU-intensive AI workloads, how to build MLOps pipelines for AI model deployment, how distributed systems stay consistent and resilient under load, how to engineer for security at scale, and how to design the LLM infrastructure that powers agentic AI applications. The work sits at the layer that decides whether an AI product survives contact with real users. It is also, crucially, not the same thing as IT. Terzi is thoughtful to draw the line. The Online MS in Computer Science & AI graduate is not the person you call when a system fails. The graduate is the person who designed the system so that, when it does fail, it fails safely, recovers automatically, and stays available to everything that depends on it.<\/p>\n The faculty teaching the program are the same tenured and senior researchers who teach Boston University\u2019s residential graduate students. That is not the norm in the online graduate market. Many programs outsource course development and instruction to adjuncts or third-party platforms. The Online MS in Computer Science & AI does not. Every course is taught by a Boston University faculty member with research credentials in the area, and several have led engineering organizations at companies like Google, IBM, and Meta before joining BU.<\/p>\n The program will also evolve in real time. Terzi described conversations with faculty who built course material three months ago and now have to redesign it because the underlying tools have changed. That cadence is unusual in higher education but is necessary in this field. The Online MS in Computer Science & AI also treats AI not as a topic to study but as a collaborator inside the learning itself. Students will use frontier models throughout the program, with explicit instruction on when, how, and why.<\/p>\n Boston University has built one of the deepest AI master\u2019s portfolios in higher education, and the Online MS in Computer Science & AI is one of five online programs in the cluster<\/a>. The temptation, with a portfolio that large, is to ask which one is best. That is the wrong question. The right question is which layer of the AI stack the prospective student wants to work at.<\/p>\n And the more interesting story is how the portfolio was built. Most universities develop their AI programs in silos, often duplicating curricula across departments and leaving prospective students to guess which version fits their goals. Boston University made a different bet. The five online programs were designed together as a coordinated portfolio, intentionally engineered to complement each other rather than compete. Each one prepares its graduates for a distinct layer of the AI economy, and the programs are built to fit how AI actually gets developed, deployed, and led in the real world. That kind of integrated design is rare in graduate education, and it gives BU graduates an unusual advantage. They emerge not just credentialed in their specialty, but oriented to the broader stack their work sits inside. Rather than developing AI programs independently, Boston University designed the portfolio as a coordinated set of offerings aimed at different layers of the AI ecosystem.<\/p>\n The Online MS in Data Science<\/a> prepares students to extract knowledge from data, to design the algorithms and analyses that turn raw signal into insight. It sits at the analytical layer.<\/p>\n The Online MS in Enterprise AI<\/a> prepares students to design, deploy, and govern AI systems at organizational scale, to operationalize AI, build resilient production systems, manage risk and governance, and translate strategy into enterprise-ready systems. It sits at the implementation and transformation layer.<\/p>\n The Online MS in Software Engineering for AI<\/a> prepares students to build the AI-powered software applications themselves. It sits in parallel with the systems work, going deep on application development.<\/p>\n The Online MS in AI in Business<\/a>, offered by Questrom School of Business, prepares business leaders to integrate AI into strategy, operations, and decision-making across functions. It sits at the business and leadership layer.<\/p>\n The working AI economy needs people at every layer of it. The Online MS in Enterprise AI graduate ships AI to production. The Online MS in Computer Science & AI graduate designed the platform that production runs on. The Online MS in Data Science graduate built the algorithm the production system serves. The Online MS in AI in Business graduate decided which business problem the AI was being asked to solve. Each role requires the others. Each program prepares its students for one of them.<\/p>\n \u201cWe build on each other,\u201d Terzi said. \u201cEverybody needs everybody.\u201d<\/p>\n The Online MS in Computer Science & AI is built for several overlapping audiences. Mid-career software engineers who studied computer science five, ten, or twenty years ago and now need to update their skills for the AI infrastructure era. Data scientists who want to move from the analytical layer into systems engineering, where the career runway extends upward into VP of Engineering and CTO roles. Recent graduates facing the junior hiring crunch, who need deeper systems knowledge than a bachelor\u2019s degree typically provides. Technical managers and program managers who need to make architectural decisions and have credible conversations with their senior engineers without losing the thread.<\/p>\n The career destinations the program is designed around are AI engineer, machine learning engineer, AI infrastructure engineer, MLOps engineer, AI platform engineer, cloud architect, site reliability engineer, senior software engineer, systems architect, VP of engineering, and eventually Chief AI Officer or CTO. These are the roles the AI economy is hiring into faster than it can fill them.<\/p>\n The rationale echoes what senior engineers told BU faculty during a recent industry panel: as AI automates more routine development work, professionals who understand the systems underneath that work become more valuable. AI can generate code, but organizations still need people who can evaluate architecture, reliability, scalability, and risk.<\/p>\n Every major AI breakthrough of the last several years has depended on infrastructure somebody designed, built, and maintained. Models train on data pipelines. Agents interact through APIs. Applications rely on distributed systems, databases, and cloud platforms.<\/p>\n The 14 hours of October 19, 2025, were a reminder that infrastructure can fail, and a reminder of who fixes it when it does. The next decade of AI will not be defined only by who builds the most capable model but also by who keeps the infrastructure underneath scalable, and safe. That work is technical. It is systems-deep. And it is one of the few categories of engineering work that the AI revolution is making more important, not less.<\/p>\nThe day the road went down<\/h2>\n
The engineers AI cannot replace<\/h2>\n
What \u201cComputer Science\u201d actually means now<\/h2>\n
Where this fits in BU Virtual\u2019s AI cluster<\/h2>\n
Who this is built for<\/h2>\n
The road has always been there<\/h2>\n