# SQL + Python Is Not Enough
For years, the formula seemed simple: learn SQL + learn Python = find data work. Especially as mid-sized companies start to become “data-driven.” Hiring managers were happy to find anyone who could write well GROUP BY then object the pandas DataFrame without breaking anything. Do you know what PostgreSQL is? Come in, you got the job! This worked for a while. Until it was.
If you haven't noticed, the job market for data professionals has undergone a structural change. yes, SQL and Python are still important; they are in every job description. But it's late reduced from separators to essentials.
Most likely, you are still preparing for the interview questions you did three years ago. Forget about it. This article addresses the gap between what candidates are preparing for and what companies currently need.
# What the Job Market Really Asks for
A January 2026 ranking by Future Proof Data Science of more than 700 data scientist jobs found that Python and SQL were still among the top three skills, but machine learning and AI skills second and fourth.
Image Source: Future Proof Data Science
Not all AI-related deployments require hands-on AI expertise, but 1 in 3 do. I some AI skills are very much needed they are:
- Major language models (LLMs)
- Recovery of advanced generation (RAG)
- Fast engineering
- Vector details
This speaks to increasing demand for data scientists who can build and implement AI systems.
Remember that the direction again the speed of this change is important. This reminds me of how machine learning went from a niche need in 2012 to near-universal in 2020.
The second story less obvious but debatable faster than most candidates: be the bar for basic engineering has been raised significantly. Data engineering skills – pipelines, orchestration, cloud platforms, data quality assessment – and machine learning in production – model monitoring, drift detection, test design – now the important things are expected instead of bonuses for posting data science jobs.
A look at any major job board confirms: along with AI skills, roles with the title “Data Scientist” are a common list Snowflake, dbt, Air flowand ETL pipeline ownership as prerequisites, not niceties.
There are four skills you may be missing. These are the new differentiators in the current job market.
# Skill #1: Data Modeling
// What It Is
Data modeling it is the ability of this design how data should be structured, related, and stored. Think of it as deciding which tables to create, what they represent, and how they relate to each other.
// What Makes a Difference
The development of tools changed the landscape. Snowflake, dbtagain BigQuery all made it easier for data scientists owns the data transformation layer. In other words, modeling decisions that used to belong to data engineers are now being delegated to data scientists.
Get the data schema wrong, and you're in dangerous waters. Often, these mistakes are not immediately apparent. When they become visible, it is too late. Your machine learning work is already affected by an engineering feature built on the wrong granular data — a direct result of a poorly modeled foundation.
// How to Find It
Take the actual dataset you're working with and redesign its schema from scratch. Ask yourself these questions:
- What organizations?
- What do they relate to?
- Which cereal makes sense?
- What questions will work most often?
After that, learn about dimensional modeling. Kimball's methoddetailed in his book The Data Warehouse Toolkit, is still a useful reference point.
# Skill #2: Improving Performance
// What It Is
Performance improvement is insightful why a query works the way it does and how to make it run faster, cheaper, or at scale. You can develop SQL queriesbut also Python pipes again performance data in general – data scientists are increasingly owning it in the end.
// What Makes a Difference
First, data volumes have grown to the extent that a good but poorly executed query can cost hundreds of dollars and production time.
Second, as mentioned earlier, data scientists must now own most of the pipeline than before. Your code should be ready for production, not just running in Jupyter notebooks.
// How to Find It
Pick a few complex SQL queries you've written, run them EXPLAIN ANALYZE them, and learn what the quiz editor actually does. Then use that to improve the question. You'll likely find at least one clue, edit, or rewrite that improves each question.
Of course a small Python pipeline, profile. There are two main tools for the time being:
- cProfile: Run with it
python -m cProfile -s cumulative your_script.pyand look at the top of the output to see the tasks that consume the most accumulated time. - line_profiler: Drills down by showing a line-by-line execution timeline within a task. Use it when cProfile tells you to which the work is slow and you need to know why.
For memoryuse it memory_profile.
Get the bottle – is it slower because the Python loop has to be vectorized? Is data loaded into memory all at once instead of in chunks? — fix, and measure the difference.
# Skill #3: Infrastructure Awareness
// What It Is
This skill means you understand system data that lives and travels. These programs include cloud platforms, distributed computing, data pipelines, storage formats, and cost models.
You must know enough about the infrastructure to design applications that can be deployed on it.
// What Makes a Difference
Also, because a good portion of the data engineer's work has fallen into the data scientist's lap. When you depend on data engineers for every infrastructure decision, you're effectively creating a bottleneck — and that's not something hiring managers want.
Infrastructure awareness includes these large interconnected areas.
You will probably need to familiarize yourself with these tools.
// How to Find It
Schedule a session with your data engineering team. Stay with them and ask them I took you to the end of the pipe. Understand where the data resides, how it is classified, and what happens if something breaks.
Then pass to build a small pipe yourself: use a free cloud tier, understand costs and usage metrics, and intentionally break down the pipeline to understand how it fails.
# Skill #4: Designing RAG Systems, Assessing LLM Results, and Assessing AI
// What It Is
This set of skills is related an active AI function. You must know how to design retrieval-augmented generation (RAG) systems (to connect LLMs to real data sources), build test frameworks (to measure that an LLM-enabled feature actually works), and perform tests on AI features.
// What Makes a Difference
AI tools are the reason. They made it possible to build a RAG pipeline without extensive research knowledge. Frameworks like LangChain again The LlamaIndexcombined with a cloud-native vector database, significantly lowers the barrier.
So the question is no longer whether it can be built – yes, it can be. But can it be properly built, tested, and trusted in production? Answering that question is what you need to be able to do: define metrics, design tests, and measure results.
In applying these skills, you will use these tools.
// How to Find It
Find out more interview questions to help you refine your AI thinking. Here are some examples from it AI Product & GenAI interview questions on StrataScratch.
Example #1: Measuring AI Feature Release in Retail Stores
How would you measure the impact of an innovative AI-powered recommendation system introduced in a sample of stores? How can you design an experiment and account for store-level variation?
Example #2: RAG System Architecture
Explain how to build a RAG system from scratch. What components are needed, and how can you improve the quality of the retrieval?
After making your thinking clear, build a small RAG application: select a domain, embed a document corpus, coordinate retrieval, and evaluate output using structured metrics.
Also, design an experiment: write a hypothesis, define metrics, and think of a valid experiment to test it.
# The conclusion
Four skills – data modeling, performance improvement, infrastructure awareness, and AI practical skills – are what bridge the gap between you and the job market. I hope you don't fall for it. To make sure you don't, this article includes practical advice on how to find each one.
Nate Rosidi he is a data scientist and product strategist. He is also an adjunct professor of statistics, and the founder of StrataScratch, a platform that helps data scientists prepare for their interviews with real interview questions from top companies. Nate writes about the latest trends in the job market, provides interview advice, shares data science projects, and covers all things SQL.
