The AI and Data Scientist trek
Classical ML to generative AI. Statistics, Python, deep learning, NLP, LLMs, RAG, computer vision, and production ML systems. The unified path for the modern AI/DS role.
Python & math foundations
The language and mathematical toolkit that everything else builds on.
Data wrangling & EDA
Cleaning, exploring, and understanding data before modeling. The work that determines whether a model is useful.
Classical machine learning
The algorithms that work well on tabular data — and the intuitions that make neural networks make sense.
Deep learning with PyTorch
Neural networks from autograd up: build CNNs and RNNs before touching pretrained models.
NLP & transformers
Text processing, embeddings, the transformer architecture, and fine-tuning pretrained models for real tasks.
Computer vision
Image classification, object detection, segmentation, and the vision models used in production.
Large language models
How modern LLMs work, what RLHF does, inference parameters, cost modeling, and prompt engineering at depth.
RAG systems
Retrieval-augmented generation end-to-end: embeddings, vector stores, chunking, reranking, and evaluation.
MLOps & model deployment
Taking models from notebooks to production: versioning, serving, monitoring, and retraining pipelines.
Experimentation & causal inference
Designing experiments that actually change decisions and going beyond correlation to understand what causes what.
AI agents & tool use
Building agentic systems that can plan, use tools, and complete multi-step tasks reliably.
Research skills & paper reading
Reading papers efficiently, implementing ideas from scratch, and staying current in a field that moves weekly.
Capstone — end-to-end AI system
Research → prototype → production → evaluation. Build something real that combines classical ML, deep learning, and LLM capabilities.
Trek complete. What's next?
You've walked the full roadmap. Now ship the capstone, write about it, and share the path with the next engineer who needs it.