I’ve been a Data Analyst for three years using mostly SQL and Tableau. I want to move into ML Engineering because the salary ceiling is much higher. What is the biggest "skill gap" I need to bridge? Is it moving from "descriptive" to "predictive" stats, or is it more about learning the software engineering side like Docker and APIs?
3 answers
The biggest shift isn't just the tools—it's the mindset. Analysts explain what happened; Engineers build systems that decide what will happen. I made this jump in 2023. For me, the hardest part was learning "Software Engineering" best practices. Writing a script for a one-time report is very different from writing a production-grade ML pipeline that needs to handle thousands of requests per second. You need to master Python (not just notebooks, but actual .py modules), Git, and how to serve a model using FastAPI. If you can take a model and wrap it in an API that a web dev can use, you are officially an ML Engineer.
Melissa, that is spot on about the "Software Engineering" part. Do you think it's worth getting a specific Cloud Certification (like AWS Machine Learning Specialty) during this transition, or is a strong GitHub portfolio more convincing to hiring managers?
Don't underestimate "Feature Engineering." As an analyst, you're already good at finding patterns. Applying that to create the right inputs for a model is your "secret weapon."
I totally agree with Diane. Justin, your analytical background means you already know how to talk to the business, which is a rare skill in the purely technical ML world. Use that to your advantage!
Gregory, a portfolio always beats a certificate in my experience. A certificate shows you can pass a test; a GitHub repo showing a deployed model with a "ReadMe" that explains your deployment choices shows you can actually do the job. If you have the time, do both, but if you have to choose, build a project that uses AWS or Azure in a real-world scenario and document it thoroughly.