We teach the things that matter.
Not just cool-looking projects: the methods, the architecture and the production systems that help you land the job and do it well. Every build is meant to be equal to real-world work, with real code and real outputs you can run yourself.
- 36 videos
- 17 runnable lessons
- 3 articles
- All code MIT licensed
What makes it different
- 01Real code, real outputEvery number on screen is printed by the code in the repo. Clone it, run it, and you get the same output.
- 02Production methodsValidation splits, step limits, typed tables, evals: the habits real systems need, not only the happy path.
- 03Architecture you can defendEach build names its parts and the hand-offs between them, so you can explain the design in an interview or a review.
Dayan Ibrar
Dayan is an AI & Machine Learning Research Engineer. He builds production data systems and applied AI, and studies what makes them reliable. His work spans data engineering, ML systems, AWS cloud, backend and APIs, full-stack and RAG, with a background in federal, aerospace and defense, and government data systems. His research projects are AXIOMERIS, TRACE and Diabeteris.
14 certifications, including
- AWS Certified Solutions Architect – Professional
- AWS Certified Generative AI Developer – Professional
- AWS Certified Machine Learning – Specialty
- AWS Certified Data Engineer – Associate
- AWS Certified Machine Learning Engineer – Associate
- Claude Certified Architect (Anthropic)
- Palantir Foundry Certified Data Engineer
What he builds with
- Languages
- PythonRustGoTypeScriptScala
- ML and AI
- PyTorchHugging FaceLangGraphMCP
- Data
- SparkKafkaPolarsDuckDBAirflowdbtRedshift
- Cloud
- AWSGCP
The series
All videos- Model AnatomyOne idea per video, from gradient descent to RAG, with the code typed and run on screen.
- Build LabProjects you can finish in a day or a weekend, file by file, like the scanned-invoices pipeline.
- Quick TipsOne trick, 20 seconds, real output.
Get in touch
For collaborations and work, reach Dayan through dayanibrar.com or the studio. For questions about a lesson, open an issue on the repo or read it as lessons on GitHub Pages.