Budget: 27000 UAH Deadline: 14 days
We can take the backend for these 3 types of tasks. Initially - 45,000 UAH and 14 days, if the current project already has ready-made lesson models, videos, transcripts, users, and interface language.
I would do this not as three separate manual endpoints, but as a normal generation and verification layer - task type, source from video text, answer options, correct answer, explanation, quality status, re-generation for weak options. Look, there’s a nuance here - for such tasks, it’s important not only to provide 3 buttons, but also to ensure that the AI does not create ambiguous answers, because then users will justifiably complain =)
> Clarification 1 - is the video text already stored in your database as a transcript, or does it also need to be obtained through AI or another service?
> Clarification 2 - should tasks be generated in advance in the admin panel or during the lesson passage by the user?
Relevant experience:
> https://business.ingello.com/fractal - AI architecture and agent scenarios, close to tasks with rule-based generation
> https://business.ingello.com/vorfahr - SaaS and automation with AI, similar logic of stable product backend
> https://systems-fl.ingello.com/ua - a brief overview of Ingello Systems and our approach to complex web systems
To start, access to the repository or backend archive is needed, a brief description of existing DRF models, and an example of one lesson with a transcript. After reviewing the code, an exact estimate can be confirmed or slightly narrowed down - based on feelings, the main risk here is indeed in the quality of AI generation, not in the APIs themselves. Overall, it’s fine, but the implementation should not be overly simplified.
