Is Udacity’s ‘Machine Learning DevOps Engineer’ for You?
It’s labeled as advanced and “advanced” is subjective. Right after this course, I jumped into Udacity’s ‘Deep Learning’—considered intermediate—and found it more challenging despite my math background.
The Ups and Downs
This course intrigued me because it combined my past & future interests in SysOps and Machine Learning. The course focuses on using tools and scripts to seamlessly deploy machine learning projects, something I was already comfortable with.
There were moments of doubt though. Luckily mentors answered almost any question I threw at them. I returned the favour by helping others – I enjoyed the time when I used my freshly acquired knowledge to resolve problems others have encountered.
Speed and Commitment
Udacity estimates around 200 hours to compIete the training. It took me roughly 250 hours of intense training in just 2.5 months.
Prerequisites
The prerequisites are basic Python and Machine Learning knowledge.
I have done Machine Learning courses on kaggle platform and I think they were sufficient to complete this nanodegree program.
The Four Projects
- Estimating Customer Churn: Easy and introductory, focusing on clean code and readability. Easy to pass if you are hands on experience with Python
| Technologies/Tools | GitHub | Pytest | Pylint | AutoPEP8 |
2. Airbnb NYC Prices: Introduced me to MLflow and Wandb, enhancing my ML toolkit. Deep dive into Machine Learning flow. Very good teacher, everything was well explained. I still use some code from that module.
| Technologies/Tools | MLFlow | Wandb | Cookiecutter |
3. Census Data: A bit skimpy on guidance, but rich in CI/CD insights. Overall a lot of knowledge that you need to find by yourself before you can code the project.
| Technologies/Tools | CI/CD | DVC | AWS S3 | Heroku | FastAPI | Workflows & pre-commits |
4. Risk Assessment: All about automating data integration and model retraining. Quite a lot of coding here. I also use some synthetic data to help me with training.
| Technologies/Tools | ML Evaluation Metrics | Crontab | Charts/Seaborn |
What I value about the course
- Varied Instruction Quality: Some modules are more thoroughly taught than others.
- Project Diversity: Expect hello of challenges that’ll test your skills from multiple angles.
I can’t recommend this course enough. The mentors, the LinkedIn and GitHub evaluations, and the progressive nature of the projects make its ROI high.
On my GitHub profile all Udacity projects are made public.
