RFP: Course Development, Online Course in Machine Learning at Scale
The Master of Information and Data Science program at the School of Information at UC Berkeley seeks proposals for a redeveloped online graduate course in Machine Learning at Scale.
About the Proposed Course
Proposals will outline a 14-week, master’s-level online learning experience that covers aspects of Machine Learning at Scale, including key concepts in parallel computation; the design of stateless parallelizable implementations of machine learning algorithms; using state of the art tools and strategies towards analysis of large scale datasets related to practical data science problems; and deploying machine learning pipelines at scale in the cloud. The successful proposal will be accepted for development and offered in the MIDS online degree program.
The existing DATASCI 261. Machine Learning at Scale course was initially developed in 2015, and has been updated several times since then. With the emergence of new deep learning frameworks, and ongoing changes in machine learning approaches, strategies, and infrastructures, the course needs to be substantially updated to remain relevant to state of the art approaches in Scalable ML. Further information about the current version of the course is available on request.
The course developer should assume that students are self-motivated, advanced master’s degree students who have completed prerequisite courses in Applied Machine Learning and Fundamentals of Data Engineering. Developers should assume students will be proficient programmers in modern programming languages (e.g., Python), have basic familiarity with deep learning frameworks (e.g., PyTorch) and architectures (such as transformers).
Topics covered in the proposed course may include, but are not limited to: scalable ML paradigms such as map-reduce, auto-differentiation, and modular architectures; implementation of machine learning algorithms in Apache Spark and PyTorch; modern machine learning algorithms, architectures, and multi-task loss functions; distributed parallel training; and deep learning techniques such as skip connections and attention.
Possible learning objectives for such a course might include:
-
Apply concepts of parallel computation
-
Design stateless parallelizable implementations of machine learning algorithms
-
Understand how Spark is used to analyze large datasets
-
Develop and deploy machine learning pipelines at scale in the cloud
-
Demonstrate proficiency in employing and specifying parameters for modular scalable structures (e.g., through PyTorch and TensorFlow)
-
Understand and use automatic differentiation techniques in machine learning deployments
We welcome input from proposers on developing course learning objectives.
The successful proposal will be accepted for development and offered in the MIDS online degree program. Since this is a fast-moving field, it is expected that the course contents will be continually revised.
Course Structure
Typically, MIDS courses have 1.5 hours of pre-recorded asynchronous lecture content per week, and 1.5 hours of live, online interactive class time per week. Classes are limited to 15–18 students, and class time typically focuses on discussion and active learning exercises focused on building knowledge and application competency.
Individual vs. Joint Proposals
The I School encourages collaborative instructional design and happily accepts joint proposals and/or proposals from individuals who would be interested in joining a teaching team. Applications should specify which component(s) of the course the individual seeks to teach and which component(s) would best be left to a co-instructor, so that we can identify developers with complementary skills. Individual proposals are also welcomed.
About the MIDS Program
The Master of Information and Data Science (MIDS) program is an innovative part-time fully online graduate professional degree program that trains data-savvy professionals and managers. The MIDS program is distinguished by its disciplinary breadth; unlike other programs that focus on advanced mathematics and modeling alone, the MIDS degree provides students insights from social science and policy research, as well as statistics, computer science and engineering.
Submission Requirements
Respondents to this RFP must submit a cover letter, C.V., and draft syllabus using the webform below.
The cover letter should include a description of the applicants’ qualifications for teaching this course including past similar courses. It should also include an assessment of whether the current course should be updated or fully overhauled. To make this assessment, the current course materials are available upon request. The current version of the course has a set of pre-recorded asynchronous lectures as well as a set of assignments. The successful applicant will be asked to review which portions should be retained or updated. We are open to partial or complete redevelopment of the course.
The draft syllabus should include:
-
A course description
-
Intended learning objectives or outcomes
-
A 14-week map of topics to be covered
-
A reading list
-
A brief description of assignments and how students will be assessed in the course (e.g., weekly homework, final project, exam, etc., along with a breakdown of grading in the course) Note: fine details about assignments are not needed until a proposal has been accepted.
-
A description of course infrastructure needs (e.g., cloud credits, platforms like Databricks, datasets, etc.).
-
A list of which course content should be retained vs. replaced.
Designers should assume some resource constraints on course structure, such as limited cloud computing and GPU availability, and are encouraged to design using pedagogical strategies that make the most of campus resources and allow for scalability.
Strong preference will be given to course developers who are interested in continuing their association with the School of Information by applying to teach the developed course as a lecturer. The separate lecturer application can be found here: https://aprecruit.berkeley.edu/JPF03780
Deadline
Responses must be received by November 14, 2023 for fullest consideration, but additional responses will be accepted until selection is complete.
Deliverables for Accepted Proposal
Course deliverables for accepted proposals will consist of well-designed, reusable presentation slides for async videos and for live classes, topic outlines and activities for classes, and assignments that exercise knowledge and skills learned in the course.
Compensation
Compensation for course development will be offered via vendor payment from UC Berkeley. To be eligible to receive compensation, the successful proposer will need to register with the UC Berkeley Accounts Payable Vendoring Team and must meet all applicable university requirements. Our expert team will walk you through the process to ensure that your vendor profile is active before work proceeds. This is not a visa opportunity.
The University of California, Berkeley is an Equal Opportunity/Affirmative Action Employer. All qualified applicants will receive consideration for employment without regard to race, color, religion, sex, sexual orientation, gender identity, national origin, disability, age, or protected veteran status. For the complete University of California nondiscrimination and affirmative action policy see: http://policy.ucop.edu/doc/4000376/NondiscrimAffirmAct
Questions
Questions about this call for proposals can be directed to review committee chair, Michael Rivera.
