You've done the Python basics. You can write loops, functions, and work with lists. Maybe you've used pandas a little. And now you're looking at the data science landscape — SQL, machine learning, statistics, cloud platforms, deep learning, AI agents — and wondering: what do I learn next?

The answer depends entirely on your goal. Someone targeting a data analyst role in Nairobi needs a different next step than someone targeting an ML engineer role in Bangalore. This post maps both paths clearly.

The "I Know Python, Now What?" Problem

Basic Python fluency is the starting point, not the finish line. The reason so many people stall after learning Python basics is that the landscape of "what to learn next" is genuinely overwhelming. There are courses on statistics, SQL, pandas, scikit-learn, TensorFlow, PyTorch, dbt, Spark, AWS, and a dozen other things — all of which someone on the internet will insist is the most important next step.

The way out of this paralysis is to anchor every learning decision to a specific job title and a specific market. Not "I want to work in data science generally" — but "I want a data analyst role at a fintech in Nairobi within 8 months" or "I want a data scientist role in Bangalore within 14 months." The target determines the stack.

The Data Career Stack — Ranked by Priority

Step
Skill
Priority
1
SQLQuerying, aggregating, joining data in relational databases. PostgreSQL or MySQL.
Must-have
2
Pandas & NumPy (deep)Data cleaning, transformation, merging dataframes, handling time series. Build 2–3 analysis projects.
Must-have
3
Data VisualizationMatplotlib, seaborn, and one dashboard tool (Metabase or Looker). Learn to communicate data visually.
Must-have
4
Statistics FundamentalsDistributions, hypothesis testing, correlation vs. causation, A/B testing basics. Not the full textbook — the decision-relevant parts.
Must-have
5
Scikit-learn (ML basics)Regression, classification, model evaluation. One real ML project on a business dataset.
High priority
6
Git & Version ControlEnough to commit, branch, and push your projects. Non-negotiable for any tech job.
High priority
7
Cloud BasicsBigQuery or AWS basics. Run a query on a large dataset in the cloud. Understand storage vs. compute.
High priority
8
LLM APIs & AI toolsOpenAI or Anthropic API basics. Build one simple AI-powered workflow. Understanding what AI can and can't do.
High priority 2026
9
Deep Learning (TensorFlow / PyTorch)Only if targeting ML engineer or research roles. Not required for most analyst or data scientist positions.
Nice to have

If Your Goal Is Data Analyst

Focus on steps 1–4 first, then step 6. This is the core data analyst stack. With SQL, pandas, visualization, and basic statistics, you are qualified for the majority of data analyst roles at companies like Safaricom, Jumia, Razorpay, Zepto, and thousands of smaller companies in Kenya and India.

Add step 5 (scikit-learn basics) once you have a solid analyst foundation. Many analyst roles now expect at least some familiarity with ML concepts even if you're not building models daily.

For analysts in Nairobi specifically: Understanding financial data patterns — mobile money flows, transaction anomalies, customer segmentation — is a huge differentiator. Learn these patterns in your projects, not just the general techniques.

If Your Goal Is Data Scientist

Work through steps 1–8 in order. The temptation is to skip steps 1–4 and go straight to machine learning. This is a mistake. Every ML interview in Hyderabad and Bangalore will test your SQL and statistics fundamentals. Models built on dirty data by someone who doesn't understand basic statistics fail in production.

Step 8 (LLM APIs) is genuinely important in 2026 even for traditional data science roles. Companies expect data scientists to understand how to work with AI tools, integrate them into pipelines, and evaluate their outputs critically.

"The data scientist who can build a machine learning model AND integrate it with an LLM for interpretation AND explain the results to a business stakeholder is worth three people who can only do one of those things."

What NOT to Learn Next

Don't learn R. Unless a specific job posting requires it, R is not the right next step for the Indian and African job markets. Python dominates data roles in these markets by a wide margin.

Don't go deep on deep learning before you have a job. TensorFlow and PyTorch are powerful and important for ML engineer roles — but they're a distraction if you're still trying to land your first analyst or general data scientist position. The majority of data science jobs do not require building neural networks from scratch.

Don't learn Spark or distributed systems yet. Spark is for processing data at scale — petabyte-level datasets across clusters. You will not encounter this problem in your first data role. Learn it when the job requires it.

Don't collect certifications instead of building projects. This applies at every stage of the learning journey, but especially here. A GitHub profile with two real analysis projects will outperform a LinkedIn profile with ten certifications in any job application in Nairobi or Hyderabad.

The Right Next Step

SQL and pandas — the highest-ROI next steps after Python basics

For most learners in India and Kenya: SQL is the highest-return next investment after Python basics. It appears in nearly every data job posting, it's fast to learn usefully, and it builds the set-based thinking that makes everything else easier.

Pick a goal (analyst or scientist). Follow the stack in order. Build projects at every stage. The path is shorter than it looks from the start.

Built for the Full Stack

Every skill in the stack, in the right order

Ramayai's tracks are sequenced exactly as described above — Python, SQL, Data Science, Machine Learning, AI Agents — with projects grounded in real East African and South Asian business problems.

🗄️ SQL 🐍 Python 🤖 AI 🎓 Placement
Explore All Tracks →
🇮🇳 Hyderabad Bangalore Mumbai 🇰🇪 Nairobi 🇳🇬 Lagos