5 Reasons I Chose Johns Hopkins Online AI Masters Program
June 30, 2024
Introduction
From 2023 to 2024, the AI market has surged from around $50 billion to $184 billion according to Statista.
Even before this growth, I started searching for AI masters programs in 2022, because I'm passionate about AI. I wanted a part-time program to avoid the opportunity cost of stopping work for two years. Since local options didn't appeal to me, I explored strong online programs.
After thorough research, I considered these online programs:
Duke's Master of Engineering in AI.
UC Berkeley's Master in Data Science (MIDS).
Johns Hopkins's Master of Science in AI.
Northwestern's Master of Science in Data Science.
I chose Johns Hopkins' AI Master of Science Program for these five reasons:
Since my work covered tuition, cost wasn't a major factor.
1. Fully Online Curriculum Aimed at Working Professionals
Johns Hopkins' AI program, part of their Engineering for Professionals programs, features small class sizes and fellow working professionals. While UC Berkeley and Northwestern also target working professionals, they require a capstone/thesis, necessitating campus visits. Duke's hybrid format worried me because in-person students might get preferential treatment over online students.
UC Berkeley's program did impress me, though, because the students there seemed to really enjoy the program and their capstone projects.
2. No GRE
None of these programs required a GRE, unlike many high-quality in-person programs.
3. Highly Ranked and Prestigious University
All the schools I considered are highly ranked. Johns Hopkins stood out for its medical network, aligning with my goal to work in AI for the medical industry.
4. No Thesis or Capstone Required
Since I don't plan to pursue a PhD, a thesis wasn't important to me. A capstone could be fun but would require taking time off work. Only Johns Hopkins doesn't require a thesis or capstone.
5. Better Electives
Johns Hopkins AI program allows you to take more electives than the other programs.
| University | Total Classes | Core Classes | Electives | Thesis/Capstone |
|---|---|---|---|---|
Johns Hopkins | 10 | 4 | 6 | No |
Northwestern | 11 | 9 | 2 | Yes |
Duke | 10 | 8 | 2 | Yes |
UC Berkeley | 8 | 4 | 4 | Yes |
I liked being able to choose more of my classes. I also didn't like some of the core classes in the other programs. For example, Duke and Northwestern require some management/business classes. UC Berkeley electives were too broad. I already work in Data Science, so I wanted more niche classes.
Johns Hopkins AI Program has more electives to choose from.
| University | Number of Electives |
|---|---|
Johns Hopkins | 35 |
Northwestern | 33 |
Duke | 19 |
UC Berkeley | 10 |
Johns Hopkins electives are more specialized.
UC Berkeley electives include:
Experiments and Causal Inference
Behind the Data: Humans and Values
Machine Learning Systems Engineering
Computer Vision
Generative AI: Foundations, Techniques, Challenges, and Opportunities
While appealing, they lacked advanced classes. For example, there were no follow-up courses in computer vision.
Johns Hopkins offers deeper specialization:
Reinforcement Learning - Introduction to Robotics - Human Robotics Interaction - UAV Systems and Control
Reasoning Under Uncertainty - Evolutionary and Swarm Intelligence - Intelligent Algorithms
Natural Language Processing - Large Language Models: Theory and Practice - Building and Training Large Language Models
Creating AI Enabled Systems (core) - Optimizing and Deploying Scalable AI Systems
Conclusion - How I've Like The Program So Far
I'm nearly halfway through the program and really enjoy the applied classes, such as applied machine learning. The theory classes are challenging but essential, especially the algorithms class, which illuminated trade-offs in algorithm design.
In my current AI class, instead of big final exams, we have weekly programming projects. This has helped me learn through practical application. For example, I've built a Streamlit app for my KNN algorithm, navigated a video game map with state space search, demonstrated a genetic algorithm for string matching, and solved a binary constraint satisfaction problem using depth-first search with smart shortcuts.
I'm particularly excited to take the machine learning engineering class to learn software engineering skills for deploying machine learning models.
In conclusion, I'm quite happy with my choice of Johns Hopkins' AI Program. It fit what I was looking for. Here's to another two years of extra busy weeks!
Last updated