🗣️ On June 4, join our next expert-led Q&A session — part of a weekly series. Ask questions and get feedback directly from course lecturers and mentors!
Get hands-on with LLM APIs and self-hosted models as you code, experiment, and build your own platform for custom AI-powered NPCs.
This 12-week course, created by experts from academia and industry, is designed specifically for developers and engineers. You’ll have a chance to get guidance from experienced pros, join regular Q&A sessions to stay on track, and connect with peers along the way.
Materials for each topic can be found in the ./topic* folders. See README.md for further details and instructions.
Any technical issues, ideas, bugs in course materials, contribution ideas - add an issue.
Study advanced concepts and apply your new knowledge right away.
Join live sessions to connect with peers and course experts, reinforcing your learning.
Basics of LLM and Multimodal LLM API usage and prompt strategies, typical problems arising with LLMs, creativity-vs-reproducibility control.
Project: Creating a chatbot and deploying it in a cloud.
LLM Workflows and beyond: from Chaining to AI Agents. LLM Reasoning.
Project: Planning and memory summarization; Automating evaluation.
RAG and its technicalities; vector stores, databases in production. RAG evaluation.
Project: Adding RAG to the NPC Factory service.
Working with open source LLMs and practical LLM inference in production. Computational and memory bottlenecks of LLM inference.
Project: Deploying a chat service based on a self-served LLM. Serving text encoders and rerankers. Making a cost-to-value choice between API and self-served LLMs.
Optimizing LLM inference, quantization and beyond. Production monitoring and observability.
Project: Optimizing open source LLM inference. Establishing monitoring with Evidently AI, Prometeus and Grafana.
Fine-tuning of LLMs and embeddings. Parameter-Efficient Fine-Tuning and LoRA. RLHF and DPO
Project: Making your characters even more alive through fine-tuning
During the course, you'll build a platform that serves intelligent, believable, reactive, and autonomous NPCs — non-player characters at the heart of immersive games, driving engaging interactions and making worlds feel rich and dynamic.
- Deploy models on real servers and integrate them into game environments.
- Implement agentic capabilities, enabling NPCs to set goals and adapt dynamically.
- Build scalable APIs to support complex game interactions.
- Optimize performance, monitor behavior, and fine-tune models for smooth, responsive gameplay.
Connect with experts, engage in discussions, ask questions, and share insights, experiences, and feedback with fellow learners on Discord. Stay in the loop — live session announcements will be posted there, too!
For updates, you can also subscribe to our newsletter.
Led by AI expert Stanislav Fedotov, this course was created by a team of AI practitioners dedicated to making AI education accessible and keeping professionals ahead in the field:
Alexey Bukhtiyarov
Nikita Pavlichenko
Sergei Petrov
Sergei Skvortsov
Alex Umnov
