Who this roadmap is for
If you already know Python at a fluent level (lists, dicts, list comprehensions, basic OOP, can read a stack trace) and want to move into AI / ML / Data Science roles within 6 months, this plan works. If you don't yet — spend 4 weeks on Python fundamentals first (the free courses page has good starting points).
This is a roadmap for the actual signal hiring managers look for — projects, fundamentals, problem-solving — not the bootcamp version optimized for course completion.
The 80/20 of what gets you hired
Out of 100 entry-level ML rejections, the cause distribution is roughly:
- ~35% — resume never reaches a human (ATS keywords, format)
- ~25% — no portfolio projects, or only Kaggle tutorials
- ~15% — can't explain ML fundamentals in interviews (bias-variance, regularization, evaluation metrics)
- ~15% — can't do basic Python coding (the LeetCode Easy/Medium tier)
- ~10% — soft skills, behavioral, communication
So 6 months should be allocated roughly: 30% projects, 25% fundamentals, 25% coding/Python, 10% resume + applications, 10% behavioral prep. The plan below maps to that.
Month 1: Math + ML fundamentals (no code-monkey trap)
The mistake: jumping straight into PyTorch tutorials without understanding linear algebra, calculus, or what a gradient actually is. You'll plateau in month 3.
Week 1-2: 3Blue1Brown's Essence of Linear Algebra (free, YouTube). Watch all 16 videos. Then his Essence of Calculus. Don't take notes — just absorb the geometric intuition.
Week 3: Andrew Ng's Machine Learning Specialization on Coursera (audit it for free). Course 1 only — supervised learning, linear/logistic regression, the math of gradient descent. Do every exercise by hand the first time.
Week 4: Read the first 4 chapters of Hands-On Machine Learning by Aurélien Géron (the second edition is online in many libraries). Code along.
Month 2: Practical ML + your first real project
Week 5-6: Finish Andrew Ng's specialization (Courses 2 + 3). Cover neural networks, decision trees, unsupervised learning, recommender systems.
Week 7-8: Build your first portfolio project — but not a Kaggle Titanic notebook. Pick a problem you actually care about. Real options that work:
- A model that predicts something about your daily life (commute time, sleep quality, food spend) from a dataset you collect.
- An end-to-end pipeline: scrape data, clean it, train a model, deploy with FastAPI to a free Render or Fly.io tier.
- A reproducible analysis of a public dataset on a topic your future employer's industry cares about.
The project must have a README, a writeup of what didn't work, and a deployed demo or a Streamlit app. "I ran sklearn on a CSV" doesn't count.
Month 3: Deep learning + your first NLP/CV project
Week 9-10: fast.ai's Practical Deep Learning for Coders (free). Lessons 1-4. The pedagogy is unusual — you train models before you learn what's inside them — and it works. Most students who do fast.ai before MIT 6.S191 say it sticks better.
Week 11-12: Second portfolio project — pick CV (image classifier on a real-world dataset you scraped) or NLP (a sentiment analyzer or topic classifier on social media data). Deploy it. Write the README like a junior ML engineer's blog post.
Month 4: LLMs / GenAI (the hiring market is here)
Half of all 2026 AI hiring is LLM-adjacent: RAG, fine-tuning, prompt engineering, evals. Knowing this stack is a massive differentiator.
Week 13-14: Andrew Ng's short courses on DeepLearning.AI (free): ChatGPT Prompt Engineering for Developers, Building Systems with the ChatGPT API, LangChain for LLM Application Development. Each is 1-1.5 hours. Do all three.
Week 15-16: Build a RAG project. Pick a real corpus (your favorite blog's archive, a textbook PDF, your company's docs if you have permission). Use OpenAI free tier or open-weights (Llama, Mistral via Ollama). Build retrieval + generation + a small eval set. Deploy it.
This is the project that gets recruiter attention in 2026.
Month 5: Python coding interviews + system design basics
You will be coded on. Don't skip this.
Week 17-18: NeetCode's Blind 75 in Python. Aim for 40-50 problems solved well, not 75 rushed. getjob4u's 100 Python questions covers the same patterns with worked solutions.
Week 19: ML system design — read "Designing Machine Learning Systems" by Chip Huyen (chapters available free in many places, also covered in her free MLOps course). Practice 3-5 mock system design questions out loud — "design YouTube recommendations", "design fraud detection for a payments company".
Week 20: Behavioral prep — write out 8 stories using the STAR method covering leadership, conflict, failure, impact, ambiguity, deadline, learning, teamwork. Practice them out loud until they sound natural.
Month 6: Resume, applications, and the actual job hunt
Week 21: Resume polish. Use a strong AI/ML sample resume as your template. Lead bullets with action verbs. Quantify every result. Cap at one page.
Week 22: Run your resume through the free ATS scanner. Hit 85+ in your target role before applying. Read the 11 ATS tweaks if you're stuck under 70.
Week 23-24: Apply at a pace of 10-15 well-targeted applications per week. JD-match your resume for each. Send 3-5 cold emails per week using the cold email generator. Track everything in a spreadsheet (date, company, role, channel, status, reply).
If you're not getting interviews after 30 well-targeted applications, the problem is the resume or the targeting — not the market. Re-scan and iterate.
What this plan deliberately skips
- Paid bootcamps — none of them teach anything you can't get from the free sources above. The $5,000-$15,000 fee buys you mostly community and accountability. Build a study group with 2-3 others doing this same plan for $0.
- Kaggle ranking grind — past Expert tier, hiring teams don't care. Two strong portfolio projects beat a Kaggle Master title for entry-level hiring.
- Master's degree — useful if you want research roles or work visas, often not worth $80K for industry ML. The OMSCS at Georgia Tech is the rare exception ($7K, online, well-respected).
- Random certifications — most don't move the needle. The ones that do are the free ones you finish: IBM Data Science, Google Data Analytics, DeepLearning.AI specializations.
Your starting point
- Bookmark getjob4u's visual career roadmaps for AI/ML/DS/Data Analyst paths.
- Pick 2-3 courses from the free courses page matching Month 1's plan.
- Block 8-12 hours/week on your calendar. Same days, same times. The single biggest predictor of success isn't IQ — it's consistency of weekly study hours over 6 months.
- Set a 90-day check-in: have you finished one shipped portfolio project by week 12? If not, you're going to overrun. Adjust scope, not deadline.
The 6-month plan works when you do the work. The plan doesn't work if you spend month 4 finishing month 1's coursework — which is the most common failure mode. Move on.