62 - Restarting my data career search
I did some research on the viability of moving towards machine learning engineering for my career. It was spurred from noticing that I was jealous of other people who were into the field. I’m tired of feeling this way, and I want to get to a place where I feel like I’m fully using my skills and capacity on the job.
I also begin to cement the knowledge that passion follows from mastery. In the past, each time I picked up machine learning I sucked at it, so I would stop. Instead, I should recognize that everybody sucks in the beginning, and that I won’t enjoy it at first because I’m not good at it yet. So having a craftsman’s mindset and continuing to put in the work is what is important and will lead me towards enjoying what I’m doing.
Derek Sivers shares a story about his early days at the Berkeley School for Music. He met an older man who taught him music theory and helped him graduate much faster than the rest of his class. The moral of the story is from the older man on the first day they met: “You don’t have to finish at the same speed as everybody else.” Similarly, I would like to improve my career more quickly than the traditional path of stitching together a couple jobs that somehow add up to being a machine learning engineer.
Before taking the leap I wanted to make sure that the job market was right for me to make this transition. I’ve been burned in the past with trying to get a data science job when it was much too competitive to do so. I looked around job websites and found a good amount of data jobs. I also took the following notes about getting into machine learning.
# Notes
Complete Roadmap for Machine Learning | ML Roadmap for Beginners
- Must be a good software engineer and understand data structures and algorithms
How to become a machine learning engineer
- For a bachelors degree, the experience required is 4 yrs
- Bay area is hottest area for ML Engineering
- Most important skill-set is machine learning
- Followed by deep learning and its fields
- Big data
- Docker
- Data analytics and visualization
How to become a machine learning engineer in 2022
- Try to get experience in data science, backend engineering, data engineering, or DevOps
- Anything you can do with data or software will get you closer to MLE
- Spend 1 year or so just programming
- Split among DS, ML, MLE, etc.
- Keep your goal in mind and spend all your time building projects
- Don’t just spend your time tuning models. Also bring them into production.
The Fastest Way to Become a Machine Learning Engineer
- Research your dream job roles first, which will give you a compass or a purpose to guide your study
How I would learn ML if I had to start over
- Have a data mindset. Think like a statistician, and then post online.
- Duke, Annie - Thinking in Bets
- SQL: window functions, analytical queries
- Data augmentation
- Applied machine learning: using machine learning to solve problems. Probably higher leverage than learning a bunch of theories at first
- Get an overview of machine learning theory: classification, regression, clustering, bias/variance, dimensionality reduction, model selection, preprocessing, basics of DNNs, ensembles, optimization, boosting, bagging, debiasing, hyperparameter tuning
- Learn with passion - enjoy what you’re learning. Do projects based on what you enjoy
- Do a deep dive and create a project from end-to-end. It’s important to finish it
- Linear algebra, statistics, and calculus
6 techniques that helped me study machine learning five days a week
- Create a curriculum. Spend up-front time to create a curriculum, then follow it
- Show up every day
- Create empty space in your day to reflect and absorb the information
I'm a freelance software developer located in Denver, Colorado. If you're
interested in working together or would just like to say hi you can reach me
at me@ this domain.