FindMyFutureAI/ML is one of the highest-paying entry points in tech — but it's also where students waste the most time learning things in the wrong order. This roadmap keeps it practical: enough maths to understand what you're doing, strong Python and ML fundamentals, then deep learning and modern GenAI/LLM skills that employers actually want in 2026.
You don't need a PhD to get an AI job. You need solid fundamentals and real projects. Follow the phases and build as you go.
Phase 1 · Foundations
Weeks 1–6
Python for data
Strong Python plus NumPy, Pandas and Matplotlib — the everyday tools of an ML engineer.
Maths that matters
Linear algebra (vectors, matrices), probability & statistics, and the intuition behind calculus/gradients. Focus on understanding, not proofs.
Data wrangling
Cleaning, transforming and exploring real datasets — 80% of the actual job is data work.
Phase 2 · Machine Learning
Weeks 7–14
Core ML algorithms
Linear/logistic regression, decision trees, random forests, SVM, k-means, KNN — how they work and when to use them.
Model workflow
Train/test split, cross-validation, overfitting, regularisation, evaluation metrics (precision, recall, F1, ROC).
Scikit-learn projects
Build end-to-end ML projects: predict, classify, cluster — with clean notebooks on GitHub.
Phase 3 · Deep Learning
Weeks 15–22
Neural networks
Perceptrons, activation functions, backpropagation, and building nets with PyTorch or TensorFlow.
CNNs & computer vision
Convolutional networks for image classification and detection.
NLP & transformers
Text processing, embeddings, RNNs, and the transformer architecture that powers modern AI.
Phase 4 · GenAI, LLMs & deployment
Weeks 23–30
LLMs & prompt engineering
How large language models work, prompting, RAG (retrieval-augmented generation), and building apps with LLM APIs.
MLOps basics
Serve a model as an API (FastAPI), containerise with Docker, and deploy to the cloud.
Capstone project
One strong, end-to-end AI project — ideally a GenAI app — deployed and documented.
Projects to build
Recruiters probe projects, they don't count them. Ship 2–3 of these, deploy them, and put them on GitHub with a clean README:
- House-price / churn prediction with scikit-learn
- Image classifier with a CNN (PyTorch/TensorFlow)
- Sentiment analysis or text classifier (NLP)
- A RAG chatbot over your own documents using an LLM API
Where this roadmap leads
Follow this path and you're prepared for roles like:
- Machine Learning Engineer
- Data Scientist
- AI / GenAI Engineer
- ML Research Assistant
Frequently asked questions
Do I need to be great at maths for AI/ML?
You need solid intuition for linear algebra, probability and statistics — not university-level proofs. Understand what the maths does; libraries handle the heavy lifting.
Is GenAI/LLM worth learning as a fresher?
Absolutely — in 2026, LLM, RAG and prompt-engineering skills are among the most in-demand and best-paid entry points into AI.
PyTorch or TensorFlow?
PyTorch is now the default in research and increasingly in industry. Learn PyTorch first; the concepts transfer.
Follow this roadmap inside the app
Personalised, trackable, with curated resources for every step.