logoNEURAL COOK
SearchAll PostsTagsNotesAuthor
logoNEURAL COOK

A posts created with Next.js and Tailwind.css

Explore

  • All posts
  • Notes
  • Snippets
  • Tags
  • RSS feed

Connect

instagramlinkedinmailgithub

© 2026 VIKI. All rights reserved.

Built with
Contents
AI SkillsMCPRoadMap
AI

RoadMap

10 min read
On this pageWeek 1: Python, NumPy and Pandas
Week 1: Python, NumPy and PandasWeek 2: Mathematics and Statistics for MLWeek 3: Supervised Machine LearningWeek 4: Advanced Classical MLDays 29–30: Classical ML ProjectPhase 1 DeliverablesWeek 5: Neural Network FundamentalsWeek 6: PyTorchWeek 7: Deep Learning ArchitecturesWeek 8: Transformers and LLM FoundationsDays 59–60: Deep Learning ProjectPhase 2 DeliverablesWeek 9: LLM EngineeringWeek 10: Advanced RAG and Fine-tuningWeek 11: MLOps and DeploymentWeek 12: Production RAG ProjectDays 89–90: Final Review and Interview Preparation

# Core Books

  1. ML: Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow — Aurélien Géron.
  2. LLM: Build a Large Language Model (From Scratch) — Sebastian Raschka.
  3. MLOps: Designing Machine Learning Systems — Chip Huyen.

# PHASE 1: Machine Learning Foundations

Days 1–30

# Week 1: Python, NumPy and Pandas

DayTopicExerciseResource
1ML overview: supervised, unsupervised and reinforcement learningIdentify 5 real-world ML problems and classify themGéron: Introduction; Google ML Crash Course
2Python for ML: lists, dictionaries, functionsWrite functions to clean and transform a datasetKaggle Python
3NumPy: arrays, dimensions, broadcastingImplement vector addition, matrix multiplication and normalizationNumPy Quickstart
4Pandas: Series, DataFrames, filtering and groupingLoad a CSV and perform 10 data manipulation operationsKaggle Pandas
5Data cleaning: missing values, duplicates and outliersClean a real-world Kaggle datasetKaggle Pandas; Géron
6Data visualization: histograms, scatter plots and correlationsPerform EDA on Titanic dataKaggle Data Visualization
7Weekly revisionCreate one EDA notebook and publish it on GitHubRevise Days 1–6

Resources:

  • https://www.kaggle.com/learn/python
  • https://www.kaggle.com/learn/pandas
  • https://www.kaggle.com/learn/data-visualization
  • https://numpy.org/doc/stable/user/quickstart.html
  • https://developers.google.com/machine-learning/crash-course

# Week 2: Mathematics and Statistics for ML

DayTopicExerciseResource
8Linear algebra: vectors and matricesImplement vector operations using NumPy3Blue1Brown Linear Algebra
9Matrix multiplication and dot productsImplement dot products and matrix multiplicationNumPy Documentation
10Probability: conditional probability and Bayes theoremSolve 10 probability problemsStatQuest
11Statistics: mean, median, variance and standard deviationCalculate statistical summaries for a datasetStatQuest
12Distributions: normal, Bernoulli and binomialVisualize three probability distributionsStatQuest
13Calculus: derivatives and gradientsImplement numerical differentiation using Python3Blue1Brown Calculus
14Optimization: gradient descentImplement gradient descent for a simple functionGoogle ML Crash Course

Resources:

  • https://www.3blue1brown.com/topics/linear-algebra
  • https://www.3blue1brown.com/topics/calculus
  • https://statquest.org/
  • https://developers.google.com/machine-learning/crash-course

# Week 3: Supervised Machine Learning

DayTopicExerciseResource
15ML workflow: features, labels, training and testingCreate your first train/test splitGéron: End-to-End ML Project
16Linear regressionImplement linear regression with Scikit-learnGéron; Kaggle Intro to ML
17Loss functions: MSE and MAECalculate both losses and compare predictionsGoogle ML Crash Course
18Logistic regressionBuild a binary classification modelGéron: Logistic Regression
19Classification metrics: accuracy, precision, recall and F1Generate a confusion matrix and calculate metricsGoogle ML Crash Course
20Overfitting, underfitting and regularizationTrain models with different complexity and compare resultsGéron
21Weekly projectBuild a Titanic survival prediction baselineKaggle Titanic Competition

Resources:

  • https://www.kaggle.com/learn/intro-to-machine-learning
  • https://www.kaggle.com/competitions/titanic
  • https://developers.google.com/machine-learning/crash-course

# Week 4: Advanced Classical ML

DayTopicExerciseResource
22Decision treesTrain a decision tree and visualize its decisionsGéron
23Random forestsCompare random forest with a decision treeGéron
24Gradient boosting and XGBoostTrain an XGBoost model on a tabular datasetXGBoost Documentation
25Feature engineeringCreate new features for Titanic datasetKaggle Feature Engineering
26Cross-validationImplement k-fold cross-validationGéron
27Hyperparameter tuningCompare GridSearchCV and RandomizedSearchCVScikit-learn Documentation
28Data leakage and pipelinesBuild a preprocessing and training pipelineKaggle Intermediate ML

Resources:

  • https://xgboost.readthedocs.io/
  • https://www.kaggle.com/learn/feature-engineering
  • https://www.kaggle.com/learn/intermediate-machine-learning
  • https://scikit-learn.org/stable/

# Days 29–30: Classical ML Project

DayTopicExerciseResource
29End-to-end ML projectComplete Titanic solution using three models, cross-validation and error analysisKaggle Titanic; Géron
30Project documentationPublish notebook with EDA, metrics, feature importance and READMEGitHub

# Phase 1 Deliverables

  • One complete Kaggle competition.
  • One polished GitHub ML notebook.
  • Understanding of classical ML algorithms.
  • Ability to explain model evaluation and overfitting.

# PHASE 2: Deep Learning and NLP

Days 31–60

# Week 5: Neural Network Fundamentals

DayTopicExerciseResource
31Introduction to neural networksDraw a neural network and explain its layersGéron
32Perceptrons and activation functionsImplement sigmoid, ReLU and tanh using NumPy3Blue1Brown Neural Networks
33Forward propagationImplement forward propagation for a small networkGéron
34BackpropagationUnderstand and manually calculate gradients for a tiny network3Blue1Brown
35Gradient descent and optimizersCompare SGD and AdamPyTorch Tutorials
36Loss functions and regularizationExperiment with different losses and dropoutGéron
37Weekly revisionImplement a small neural network using NumPyRevision

Resources:

  • https://www.3blue1brown.com/topics/neural-networks
  • https://pytorch.org/tutorials/

# Week 6: PyTorch

DayTopicExerciseResource
38PyTorch tensorsCreate tensors and practice reshaping, slicing and broadcastingPyTorch Basics
39Autograd and computational graphsCalculate gradients using PyTorch autogradPyTorch Autograd
40Building neural networks using nn.ModuleCreate a two-layer neural networkPyTorch Tutorials
41Training loopsImplement forward pass, loss, backward pass and optimizer stepPyTorch Tutorials
42Dataset and DataLoaderLoad and batch a custom datasetPyTorch Tutorials
43GPU training and optimizationTrain on GPU if available; compare CPU and GPUPyTorch Tutorials
44Deep learning mini-projectBuild an MNIST digit classifier using PyTorchPyTorch MNIST Example

Resources:

  • https://pytorch.org/tutorials/beginner/basics/intro.html
  • https://pytorch.org/tutorials/

# Week 7: Deep Learning Architectures

DayTopicExerciseResource
45CNN fundamentalsUnderstand convolution, filters and poolingGéron
46CNN implementationBuild an image classifier using PyTorchPyTorch Computer Vision
47RNN and LSTM fundamentalsDraw an RNN architecture and understand sequential processingUnderstanding LSTM Networks
48NLP fundamentalsClean text and build a preprocessing pipelineHugging Face NLP Course
49TokenizationImplement word-level and subword tokenization experimentsHugging Face Tokenizers
50Word embeddingsGenerate embeddings and calculate cosine similaritySentence Transformers
51Text classification projectBuild a sentiment classifier using TF-IDF and logistic regressionKaggle NLP datasets

Resources:

  • https://colah.github.io/posts/2015-08-Understanding-LSTMs/
  • https://huggingface.co/learn/nlp-course/
  • https://www.sbert.net/

# Week 8: Transformers and LLM Foundations

DayTopicExerciseResource
52Attention mechanismImplement scaled dot-product attention in PyTorchRaschka
53Self-attentionCalculate query, key and value representationsRaschka
54Multi-head attentionImplement a multi-head attention moduleRaschka
55Positional embeddingsImplement token and positional embeddingsRaschka
56Transformer architectureDraw and explain a Transformer blockIllustrated Transformer
57GPT architectureUnderstand decoder-only Transformer architectureRaschka
58Tokenization and language modelingExplore BPE tokenization and next-token predictionHugging Face LLM Course

Resources:

  • https://jalammar.github.io/illustrated-transformer/
  • https://huggingface.co/learn/llm-course/
  • https://github.com/rasbt/LLMs-from-scratch

# Days 59–60: Deep Learning Project

DayTopicExerciseResource
59Fine-tuning pretrained modelsFine-tune DistilBERT for sentiment classificationHugging Face Transformers
60Model evaluation and publishingEvaluate precision, recall and F1; publish model and model cardHugging Face Model Hub

# Phase 2 Deliverables

  • A neural network implemented with PyTorch.
  • An image classification exercise.
  • A sentiment classification model.
  • Understanding of attention and Transformer architecture.
  • One model published on Hugging Face.

# PHASE 3: Generative AI, MLOps and Production Engineering

Days 61–90

# Week 9: LLM Engineering

DayTopicExerciseResource
61LLM fundamentals and inferenceUnderstand pretraining, inference and next-token predictionRaschka
62Prompt engineeringExperiment with zero-shot, few-shot and structured promptsPrompt Engineering Guide
63Hugging Face Transformers pipelineLoad a pretrained model and perform inferenceTransformers Documentation
64Embeddings and semantic searchGenerate sentence embeddings and retrieve similar sentencesSentence Transformers
65Vector databasesBuild a semantic search system using FAISSFAISS Documentation
66Chunking and document preprocessingLoad PDFs, split documents and experiment with chunk sizesLlamaIndex Documentation
67Retrieval-Augmented GenerationBuild a basic RAG pipeline using retrieved document chunksHugging Face; RAG Tutorials

Resources:

  • https://huggingface.co/docs/transformers/
  • https://www.promptingguide.ai/
  • https://www.sbert.net/
  • https://github.com/facebookresearch/faiss
  • https://docs.llamaindex.ai/

# Week 10: Advanced RAG and Fine-tuning

DayTopicExerciseResource
68RAG architectureDraw complete ingestion and retrieval architectureChip Huyen; RAG documentation
69Retrieval qualityExperiment with top-k retrieval and similarity thresholdsLlamaIndex
70RerankingCompare similarity search with a cross-encoder rerankerSentence Transformers
71RAG evaluationEvaluate retrieval precision, recall and answer relevanceRagas Documentation
72Hallucination and grounded answersCreate 20 test questions and identify unsupported answersRagas; Hugging Face
73Fine-tuning vs RAGCompare when to use RAG, prompting and fine-tuningRaschka; Hugging Face
74LoRA and PEFTFine-tune a small pretrained model using parameter-efficient fine-tuningPEFT Documentation

Resources:

  • https://docs.ragas.io/
  • https://huggingface.co/docs/peft/
  • https://docs.llamaindex.ai/

# Week 11: MLOps and Deployment

DayTopicExerciseResource
75ML lifecycle and production architectureDraw a complete model development-to-production pipelineChip Huyen
76Experiment trackingTrack model parameters and metrics using MLflowMLflow Documentation
77Model versioningSave model artifacts and track model versionsMLflow Model Registry
78FastAPIBuild an API endpoint that accepts input and returns model predictionsFastAPI Tutorial
79DockerContainerize your ML applicationDocker Get Started
80Model serving and inferenceServe your trained model through FastAPI and DockerChip Huyen
81Monitoring and driftSimulate changing input distributions and monitor model metricsChip Huyen

Resources:

  • https://mlflow.org/docs/latest/
  • https://fastapi.tiangolo.com/tutorial/
  • https://docs.docker.com/get-started/

# Week 12: Production RAG Project

DayTopicExerciseResource
82Project architectureDesign an end-to-end document question-answering applicationFull Stack Deep Learning
83Document ingestionUpload and parse PDF documentsLlamaIndex
84Embeddings and vector storageGenerate embeddings and store them in FAISSSentence Transformers; FAISS
85Retrieval pipelineImplement top-k retrieval and rerankingLlamaIndex; Sentence Transformers
86LLM generationGenerate answers using retrieved contextHugging Face or an LLM API
87Evaluation and testingCreate a test dataset and evaluate retrieval and answer qualityRagas
88DeploymentDeploy your RAG application using FastAPI and DockerFastAPI; Docker

# Days 89–90: Final Review and Interview Preparation

DayTopicExerciseResource
89ML system design and interview preparationDesign a recommendation system and scalable RAG system; explain architecture and trade-offsChip Huyen; ML System Design
90Portfolio and career readinessPolish GitHub, READMEs, project architecture diagrams and resume; prepare project walkthroughsGitHub
←PreviousMCP

On this page

Week 1: Python, NumPy and PandasWeek 2: Mathematics and Statistics for MLWeek 3: Supervised Machine LearningWeek 4: Advanced Classical MLDays 29–30: Classical ML ProjectPhase 1 DeliverablesWeek 5: Neural Network FundamentalsWeek 6: PyTorchWeek 7: Deep Learning ArchitecturesWeek 8: Transformers and LLM FoundationsDays 59–60: Deep Learning ProjectPhase 2 DeliverablesWeek 9: LLM EngineeringWeek 10: Advanced RAG and Fine-tuningWeek 11: MLOps and DeploymentWeek 12: Production RAG ProjectDays 89–90: Final Review and Interview Preparation