AI foundations and modern tools
From everyday AI use to an engineering workspace: model landscape, Codex, Git and GitHub, agent harnesses, instructions, skills, memory, and safe work in small verified cycles.
Agentic AI & LLM: from foundations to autonomous systems
Learn by building. The portal organizes public course recordings into a clear route with source materials on GitHub.
29 video lessons12 modules35:03:00 video120 academic hours
Now playing
Module 1: AI foundations and modern tools · Part 1
1:36:50 · Learn by building. The portal organizes public course recordings into a clear route with source materials on GitHub.
Full curriculum
Not a tool list—a path from a problem to a working, accepted system.
4 video lessons
Large language models
How LLMs turn tokens into output; context windows, temperature, structured input and output, APIs, cost, hallucinations, evaluation, and choosing a model for a concrete task.
Building practical AI applications
Three practical application patterns. Start from the business task and real inputs, define the output contract, then build and test a narrow end-to-end slice before polishing the interface.
Cloud infrastructure and agent systems
VPS, Linux, SSH keys, DNS, domains, processes, ports, reverse proxy, HTTPS, deployment, logs, secrets, backups, and the difference between a local demo and an operated service.
AI-assisted web development
AI-assisted web development from PRD to production: information architecture, frontend and backend boundaries, databases and APIs, Git branches and pull requests, testing, accessibility, and deployment.
Agent data, memory, and context
How agents retain and retrieve information: files, relational data, embeddings, vector search, working context, durable memory, provenance, access boundaries, and when not to use RAG.
Business process automation
Business automation across CRM, SMM, documents, APIs and background work. Webhooks, cron, systemd and workflow tools are combined with ownership, retries, audit, human approval, and measurable acceptance.
Voice, speech, and telephony
Speech-to-text, text-to-speech, streaming audio, multilingual quality, latency, telephony, signed webhooks, consent and recording, conversation design, and post-call analysis.
Vision, image, and video
Image and video generation as a controlled pipeline: briefs, references, rights, reusable templates, brand consistency, Remotion and FFmpeg assembly, review stages, and publication preparation.
Agentic systems in the real world
Agentic systems inside real organizations: tools, permissions, memory and autonomous schedules; multi-agent coordination; automation versus transformation; hybrid teams, economics, safety, and operating-model change.
Project architecture and delivery
Turn the student project into a coherent product: architecture and threat boundaries, repository structure, runtime, data migration, observability, tests, documentation, second-operator readiness, and a defensible demo.
Project defense
Project defense against the original task: show the working route, evidence, limitations, economics, risks, ownership and next gate. A launch without business acceptance is not treated as completion.
How the 120 academic hours work
The videos are class recordings, not the whole program.
35:03public recordings
48 acad / 36 hindependent project work
12 × 10academic hours per module
120 / 90 hacademic / clock-hour total
What you will be able to build
LLM applications with reliable inputs, outputs, and evaluation
Automated workflows connected to APIs, documents, CRM, and business systems
Voice, vision, and multimodal agent interfaces
Production-ready AI systems with infrastructure, access controls, and human approval
Code, presentations, and assignments
The open GitHub repository contains the current program, Canvas presentations, PDFs, practical tasks, and archived course streams.
Agentic AI course
29 video lessons · 12 modules