Overview (Problem + Context)
- Students used fragmented tools for notes, planning, and practice tests.
- Existing study apps lacked personalization and feedback loops tied to performance.
- The objective was to create a single workflow that improves consistency and confidence.
My Role + Responsibilities
- Scoped MVP, authored product requirements, and prioritized feature rollout.
- Implemented backend services, prompt orchestration, and front-end interfaces.
- Defined instrumentation events to measure adoption and learning behavior.
Approach
- Interviewed student users to identify high-friction points in revision routines.
- Implemented retrieval pipeline over uploaded class notes and syllabus content.
- Generated adaptive quizzes based on confidence and historical accuracy.
- Added spaced-review reminders and progress checkpoints to support long-term retention.
Implementation Details (Stack + Architecture Choices)
- Built application with Next.js App Router and server actions for streamlined data mutations.
- Used PostgreSQL and Prisma for user progress, quiz history, and content indexing.
- Integrated OpenAI API for summarization, question generation, and conversational tutoring.
- Structured prompts for deterministic output format to support UI reliability.
Results (Metrics + Outcomes)
- Reached 37% week-4 retention during pilot with university students.
- Increased average weekly study session count from 2.1 to 3.8.
- User feedback highlighted stronger confidence before exams and better topic prioritization.
What I’d Do Next (Reflection)
- Add offline mode and sync to support intermittent connectivity.
- Introduce instructor dashboards for cohort-level learning insights.
- Expand personalization with mastery models and confidence calibration.
Links
- [GitHub repository](https://github.com/your-github/ai-study-coach)
- [Live prototype](https://example.com/ai-study-coach)
- [Case study report](https://example.com/ai-study-coach-report)