This is the English companion to our long-running Bangla FAQ. The tone and answers follow the original; links point to English pages where they exist.
1. Where do I start with machine learning?
Start with the Resources page — the book order, free courses and playlists live there. If you read Bangla, the Hatekolome Machine Learning series is the most beginner-friendly path.
2. Do I need a strong math background?
No PhD required. You need comfort with basic algebra, some statistics and probability — enough to understand what models are doing, not to reinvent them. Khan Academy is excellent for the basics. Our books teach the math in context.
3. Which programming language — R or Python?
Either. R is a gift for non-programmers coming from statistics; Python dominates industry and deep learning. My first book uses R; From Zero to Python Machine Learning uses Python. Pick one and finish a project.
4. Do I need a computer science degree?
No. Many successful data analysts and scientists come from other fields. What matters is skill you can demonstrate — projects, portfolios, and the ability to solve a real problem.
5. How long until I can work professionally?
With focused practice (roughly 10–20 hours a week), analytics skills in 5–6 months and a fuller data-science track in 10–16 months is realistic. See the Data Science Roadmap.
6. Which tools should I learn first?
Spreadsheets (Excel/Google Sheets), then a programming environment (R or Python via Anaconda), then SQL, then visualization (Matplotlib, Seaborn, Tableau or Power BI). Tools follow process.
7. Is machine learning just ChatGPT or image generators?
No. Those are visible tips of a deep field: predictive models, recommendations, fraud detection, forecasting, NLP, computer vision and much more. Generative AI is one branch.
8. Will AI take my jobs?
AI changes jobs more than it erases them. Roles that pair domain knowledge with data skills grow; pure routine shrinks. Reskilling is the answer — see question 26.
9. What about free online courses vs paid ones?
Google, IBM and Microsoft certificates on Coursera are high quality; free audit modes exist. Paid tracks add structure and financial-aid paths. Free content without accountability is harder to finish — that is why mentorship programs exist.
10. Which certificate should I start with?
The Google Data Analytics Professional Certificate is the best entry point for non-programmers today. IBM and Microsoft alternatives are strong too.
11. Do certificates matter to employers?
As screening signals, yes — especially early on. What gets you hired is solving the company’s problem; certificates open the door faster when many applicants exist.
12. How do I build a portfolio?
Do end-to-end projects (Iris, Titanic, Superstore, recommendation systems), write them up clearly, publish code on GitHub Pages or GitHub, and connect each project to a business question. Showcasing is half the work.
13. What is exploratory data analysis (EDA)?
EDA is exploring, cleaning and visualizing data before modeling — understanding distributions, outliers and patterns with pandas, NumPy, Matplotlib and Seaborn.
14. Why is SQL important?
Data lives in databases. SQL gets it out and aggregates it. DDL/DML basics, joins and unions are enough to start; MySQL is a fine open-source start.
15. What about NoSQL?
Optional. MongoDB is the usual example for non-tabular data. Learn it when your data demands it.
16. Do I need deep learning?
Not at the start. Classical machine learning solves most business problems. Deep learning matters for images, speech, language and large-scale pattern problems — after clean data and strong basics.
17. Which deep learning framework?
TensorFlow is the common production choice; PyTorch is preferred in research. My advanced book Hands-On Python Deep Learning covers TensorFlow.
18. How do I deploy models?
Learn the pipeline idea first (collection → storage → cleaning → analysis → modeling). Deployment, APIs (JSON, REST, OpenAPI) and basic data engineering come next — optional in the roadmap but valuable.
19. What about A/B testing and experimentation?
Core analytics skill. After solid EDA and basic models, A/B frameworks and simple algorithms (e.g. linear regression for behavior prediction) help product and business decisions.
20. Is R dying?
No. R remains excellent for statistics and visualization; Python leads industry ML. Many practitioners use both.
21. Should I learn Java/C++?
Not to start. Python/R cover the analytics path. Systems languages help later for performance-critical engineering.
22. How do I stay updated?
Follow quality documentation (RTFM — the manual the builder wrote), a few newsletters, and communities — not every hype cycle.
23. Where are the video playlists?
On Facebook page → “Videos” tab, and on YouTube as separate playlists. From the Facebook page, “Learn More” reaches the YouTube links. Channel: YouTube @aiwithr.
24. What about your current job situation?
E-commerce and service-industry leaders often ask me to find “good” data scientists; few can be referred because industry use-cases and our data scientists do not yet meet at that level. That is a market gap, not their fault. Those who can define their own industry’s problem with ML learn fastest — challenge yourself where you already work, then switch into pure data-science companies when ready.
25. There are no jobs in Bangladesh — what do I do?
Legitimate concern. The few ML jobs demand more experience than many expect; frustration is real. But careers now need multiple skillsets over a lifetime — I have re-skilled 7 times in a 28-year career. Certificates matter less than demonstrating you can solve the specific problem. Think global (Europe, Japan and others recruit data skills); learn from wherever you are.
26. What is your career advice in one piece?
Treat a skillset as life-and-death only in the sense of staying current — update tools continuously. Companies hire for specific problems; a GitHub profile and direct problem conversations beat social profiles. Reskill when the market moves. The market is emerging: “Who knows when lightning could strike?”
27. How do I learn deep learning?
A long answer lives in the Bangla article Deep Learning and ML. Start only after solid classical ML and Python.
28. Where is hands-on Bangla NLP?
Four NLP chapters live in Hands-On Deep Learning; another book is in progress: nlp_bangla on GitHub.
29. Will you write a book on ML math and algorithms?
My focus is practical applied work. Academics write the math and algorithm books; my job is applying technology correctly to help people. A little math is still inside the books — but not hardcore theory for its own sake.
30. How do I reach you in an emergency?
Machine learning is my passion, but I do it outside office hours, work and family. First read the answers on this page, then Messenger Rakibul Hassan or 01713095767 (5–6 pm).
Thank you all.