Understand this story
On November 9, 2026, an interview released by Latent Space showed that the second session of Stanford CS146S, taught by Mihail Eric, brought open-source maintainers into the course. Students must learn unfamiliar codebases, propose changes, and explain their solutions to maintainers who may merge the patches. The course lasts ten weeks, and its origin was that a single prompt to a coding agent could complete last year's assignments. There are currently no subsequent developments.
AI-generated from public reports
Latest development · Report
Latent Space interviews Stanford CS146S instructor Mihail Eric: What should engineers learn in the AI era?Developments
1 confirmed developments
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Nov 9
Shao Meng
Latent Space interviews Stanford CS146S instructor Mihail Eric: What should engineers learn in the AI era?
Latent Space invited Mihail Eric, instructor of Stanford CS146S "The Modern Software Developer" and head of Monaco AI, to discuss the direction of engineer training in the AI era. He mentioned that in a classroom survey, nearly all surveyed students had already used coding Agents, so the new round of the course incorporated real open-source contributions and maintainer review into assignments, shifting the assessment focus from whether one can generate code to whether one can propose reasonable changes and take responsibility for them. In Monaco's practice, code review is procured through Devin, while the Agent workspace is built in-house as Monacoder based on Coder, and a Slack bot named Hal was built to serve different roles such as engineers and sales.
Latent Space
How Stanford CS146S Teaches AI-Native Engineering: An Interview with Mihail Eric and Monaco
The second installment of Stanford CS146S, taught by Mihail Eric, brings open source maintainers into the course: students must learn unfamiliar codebases, propose changes, and explain their solutions to maintainers who may merge the patches. The course runs for ten weeks, and the motivation is that a single prompt to a coding agent can now complete last year's assignments.