# Systems and Toolchains for AI Engineers **Program:** MS, AI in Engineering **Units / Format:** 12 units, 2 sessions/week, in-person, hands-on **Prerequisites:** Basic programming skills (Python preferred). No prior ML expertise required. ## Class The course will be Mondays and Wednesdays from 3:30pm to 4:50pm in [CIC 1201](https://maps.app.goo.gl/eiN6HQujptAP8yoB9). Attendance in person is expected. The course is run on [Canvas](https://canvas.cmu.edu/courses/54976). ## Instructors - John Kitchin (jkitchin@andrew.cmu.edu) - Victor Alves (victoralves@cmu.edu ) ## Teaching assistants - Tirtha Vinchurkar (tvinchur@andrew.cmu.edu) - Roberto Jimenez (robertoj@andrew.cmu.edu) - Nicolas Smits (nsmits@andrew.cmu.edu) - Yiyin Zhang (yiyinz@andrew.cmu.edu) - Peter Cook (petercoo@andrew.cmu.edu) --- ## Course Description Building AI in an engineering setting is far more than choosing a model. It is an engineering discipline in its own right: 1. Acquiring and storing messy sensor, simulation, and experimental data. 2. Building reproducible pipelines; training and rigorously evaluating models. 3. Adapting ML/AI models and wiring up LLM agents, and 4. Deploying and monitoring all of it in real production. This course gives students **hands-on experience** with the **systems and toolchains** an AI engineer uses, considering **engineering problems**: time-series and sensor data, simulation surrogates (ML models) and so on. The course is balanced across three main topics/arcs: 1. **Data infrastructure & engineering:** storage, pipelines, features, validation. 2. **Machine learning:** the ML workflow, PyTorch/JAX, and applied engineering ML (surrogates, physics-informed models, uncertainty, Bayesian optimization (BO)). 3. **LLM & agentic engineering:** usage, retrieval, adaptation, and building/using agentic workflows. ## Learning Objectives By the end of the course, students will be able to: - Create a **reproducible AI project** (environments, versioning of code/data/models, experiment tracking) and defend engineering design choices made. - Acquire, store, and process engineering data across different types of databases; build **validated data pipelines**. - Execute the full **ML workflow** (training, cross-validation, model selection, and evaluation) on engineering data. - Build **applied-engineering ML**: surrogate and physics-informed models with **uncertainty quantification**, and use **Bayesian optimization / active learning** for solving problems. - Use **LLMs** effectively (prompting, structured output, embeddings), using **RAG**. - **Deploy, monitor, and operate** AI systems, and reason about safety, cost, and responsible use in an engineering context. ## The Toolchain (standardized) To keep the course coherent, the following tools are standardized across materials, and alternatives are named where relevant so the skills you learned can transfer. - **Language & environments:** Python, managed with **uv** (reproducible, lockfile-based). - **Machine learning:** **PyTorch and/or JAX**. - **Experiment tracking:** **MLflow, MLOps**. - **Data:** SQL and similar. - **LLM / agentic:** framework- and provider-**agnostic**. Students may choose their agent framework. Main concepts and ideas are taught so they transfer across model providers. ## Assessment | Component | Weight | |--------------------------------------|--------| | Module assignments | 25% | | Mini-project | 20% | | Final project (open, student-chosen) | 40% | | Participation & quizzes | 15% | There is **no proctored final exam**, and assessment is based on projects, quizzes, and homework. - **Assignments:** one assignment/homework per module, each reinforcing that week's tools. Submitted via Canvas. Typically due one week after release. - **Mini-project:** a team project, in groups of about four working in pairs the first week, that builds and compares two unsupervised fault detectors for a chemical plant, trained on normal operation only, and ends in a short report. Released with Lecture 9, due before fall break. - **Final project:** **student-chosen**. Must integrate remaining topics: an **LLM/agentic system** with real **evaluation and deployment**. Deliverables: proposal, build (repo), and presentation. - **Participation & quizzes:** one short practice module per week, tied to that week's lectures and notes. Completing a module produces a PDF that you upload to Canvas. ### Grading scale Grades are absolute, and no curve is applied: | Grade | Minimum score | |-------|---------------| | A+ | 100 | | A | 95 | | A- | 90 | | B+ | 85 | | B | 80 | | B- | 75 | | C | 60 | ## Policies - **Late work:** **All students** get two grace days from the assignment deadline, **automatically, no questions asked**. A zero will be assigned after this. Please note this only applies to the participation and assignments. **It will not be applied to the projects (group assignments).** Exceptions must be approved in advance by the instructors. - **Regrade requests:** bring a regrade request to the instructors within one week of the grade being returned. We will review it with the grader who graded the work. - **Collaboration:** working together is allowed and encouraged on all coursework, the assignments, the miniproject, and the final project alike. Disclose your collaborators and any outside sources in your repository, and make sure you personally understand and can defend everything you submit. - **Generative-AI use:** *permitted and encouraged as an engineering tool*: this is a course about building with AI. We ask you to (a) **disclose** where and how you used AI assistants, (b) **cite** generated code/text in comments or a `CREDITS` file, and (c) be able to **explain and defend** everything you submit. Using AI to bypass learning an assignment's core skill (e.g., having it write an entire graded pipeline you can't explain) is not permitted. When in doubt, disclose. - **Academic integrity:** governed by [CMU's guidelines](https://www.cmu.edu/policies/student-and-student-life/academic-integrity.html). ## Other Policies and Procedures **Accommodations for Students with Disabilities:** If you have a disability and have an accommodations approval from the Disability Resources office, we encourage you to discuss your accommodations and needs with us as early in the semester as possible. We will work with you to ensure that the appropriate accommodations are provided. If you suspect that you may have a disability and would benefit from accommodations but are not yet registered with the Office of Disability Resources, we encourage you to follow the online procedures for obtaining accommodations at . **Statement of Support for Students' Health and Well-Being:** We understand that, at times, we may be under increased stress and challenges, and we will do our best to be accommodating. All of us benefit from support during stressful times, and we will do our best to help if you choose to come to us with any concerns. There are many helpful resources available on campus, and an important part of the college experience is learning how to ask for help. If you or anyone you know experiences any academic stress, difficult life events, or feelings of anxiety or depression, we strongly encourage you to seek support. Counseling and Psychological Services (CaPS) is here to help: call 412-268-2922 and visit their website at . Consider reaching out to a friend, faculty, or family member you trust for help getting connected to the support that can help. If you are feeling desperate, call the Re:solve Crisis Network at 888-796-8226. If the situation is life-threatening, call the police: on campus, CMU Police at 412-268-2323; off campus, 911. **Statement of Support for Our Diverse Campus Community:** We are committed to treating every individual with respect, and we expect Carnegie Mellon University students to do the same. We personally recognize that we are diverse in many ways, and we believe that this diversity is reflective of society and brings value to our campus. We agree with the university's statement that diversity can refer to multiple ways that we identify ourselves, including but not limited to race, color, national origin, language, sex, disability, age, sexual orientation, gender identity, religion, creed, ancestry, belief, veteran status, or genetic information. We are personally committed to ensuring that Carnegie Mellon provides an inclusive and welcoming environment. If you feel that you have not been treated with respect in this course, we encourage you to express that, and we hope you will feel comfortable contacting us directly. If you do not feel comfortable, we would encourage you to reach out to our Chemical Engineering Department Head, Professor Carl Laird, and request anonymity if you prefer. We encourage anyone who experiences or observes unfair or hostile treatment on the basis of identity to speak out for justice and support. Anyone can share these experiences using the following resources: - Center for Student Diversity and Inclusion: csdi@andrew.cmu.edu, (412) 268-2150. - Report-It online anonymous reporting platform: (username: tartans, password: plaid). All reports will be documented and deliberated to determine if there should be any following actions. Regardless of incident type, the university will use all shared experiences to transform our campus climate to be more equitable and just. ## Schedule See [the schedule](./schedule.md) for the full tentative calendar for our course! :)