yandexdataschool/nlp_course

What it solves

This project provides a comprehensive, structured curriculum for learning Natural Language Processing (NLP). It bridges the gap between foundational concepts like word embeddings and advanced modern topics like Large Language Models (LLMs), agents, and production systems.

How it works

The course is organized into 14 weekly modules, each containing a mix of lecture materials, interactive seminars, and hands-on homework assignments. Students progress from building basic language models from scratch to fine-tuning pre-trained models, implementing RAG architectures, and deploying AI agents.

Who it’s for

It is designed for students and practitioners who want to build a deep technical understanding of NLP, ranging from the mathematical foundations of word vectors to the practical engineering challenges of deploying LLM-based systems.

Highlights

• Covers the full spectrum of NLP, from classic count-based methods to modern LLM techniques.
• Includes practical modules on PEFT (LoRA), RLHF, quantization, and distillation.
• Features dedicated sections on AI agents, multimodal models, and building LLM systems for production.
• Provides interactive lecture materials and coding seminars for every topic.
• Developed by a collaborative team of experts and refined by over 30 contributors.