Beaver Study Orchestrator

• 

  Main Service

## Inspiration

Students do not usually fail because they lack motivation. They fail because course requirements are spread across messy syllabus text, and planning becomes a last-minute guess. We wanted to build a tool that turns unstructured academic information into a daily execution system, not just another “to-do list” app.

Our goal was simple: from raw syllabus text, produce a realistic plan, explain deadline risk, and give students a clear next action before stress compounds.

## What it does

Beaver Study Orchestrator converts free-form syllabus text into an operational study workflow:

• Extracts dated tasks (assignments, labs, projects, exams) from raw text
• Estimates workload and builds an adaptive day-by-day plan from weekly availability
• Calculates an interpretable deadline risk score with top contributing drivers
• Runs a what-if simulation (for example, +1h/day) to show projected risk reduction
• Exports schedule output as .ics so users can execute in Google/Apple Calendar

## How we built it

We built a deterministic full-stack app focused on speed, explainability, and demo reliability:

• Backend: FastAPI with modular core services (syllabus_parser, scheduler, risk_model, calendar_export)
• Frontend: single-page Vanilla JS + HTML/CSS for low-friction demo flow
• Testing: Pytest unit/API tests plus CI on push/PR

Risk modeling is intentionally interpretable rather than black-box. A simplified form is:

[ \text{risk} = \sigma \left( w_1 \cdot \text{coverage_gap} + w_2 \cdot \text{urgency} + w_3 \cdot \text{workload_pressure} + w_4 \cdot \text{capacity_strain} \right) ]

This helped us show why risk is high, not only that it is high.

## Challenges we ran into

• Parsing real-world syllabus language was noisy and inconsistent (mixed formats, partial dates, vague lines)
• Capacity constraints made naive plans look “complete” when they were actually infeasible
• We had to prevent failure states from breaking the demo (for example, no valid due dates found)
• Balancing speed and transparency required careful API/UI design so users could trust the output quickly

## Accomplishments that we're proud of

• Built an end-to-end workflow from text ingestion to calendar export
• Delivered interpretable risk analytics with actionable mitigation guidance
• Added what-if simulation to support decision-making, not just reporting
• Implemented honest spillover reporting when workload exceeds available time
• Shipped with tests and CI, making the project portfolio/interview ready

## What we learned

• In productivity tools, reliability and clarity matter more than model complexity
• Users trust systems that expose constraints and reasoning
• “Execution-ready output” (calendar events, daily workload) creates more value than static analysis
• Deterministic pipelines can be a strong choice when reproducibility and cost control are priorities

## What's next for Beaver Study Orchestrator

• Add Google Calendar and LMS (Canvas) direct integration
• Personalize effort estimates using user completion history
• Support multi-course conflict optimization and priority weighting
• Add reminders and progress feedback loops for long-term habit retention
• Expand parser coverage for broader syllabus styles and multilingual inputs

Built With

• cloudflare
• fastapi
• github-action
• html/css
• icalendar
• javascript
• pages
• pydantic
• pytest
• python
• uvicorn