assist+

assist+ instantly analyzes your transcript against UC transfer requirements. Upload, select your major, and see exactly what courses youve completed and what you still need. Transfer planning, solved.

Comment

Inspiration

Every year, over 100,000 California community college students attempt to transfer to a UC campus. The process should be straightforward—but it's not. Students spend countless hours on ASSIST.org, manually cross-referencing their completed courses against thousands of articulation agreements, trying to figure out if they're on track. We've been there. The anxiety of wondering "Did I take the right classes?" The spreadsheets. The confusion when course names don't quite match. The panic when you realize you might have missed a requirement. We built assist+ because we believe transfer planning shouldn't require a PhD in bureaucracy. Every student deserves to know exactly where they stand—instantly.

What it does

assist+ transforms the transfer planning experience from hours of manual research into a 30-second upload. Upload your transcript – Just drop in your unofficial transcript PDF Select your goals – Choose your target UC campus and intended major Get instant clarity – See exactly which transfer requirements you've completed, what's still needed, and your overall progress percentage

How we built it

Frontend: React with Vite for a fast, responsive UI. We used Tailwind CSS for styling and React Router for navigation. Backend: Flask API server handling transcript analysis, agreement matching, and recommendation generation. AI Integration: We use OpenRouter's API with Google's Gemini 2.0 Flash model for two key tasks: Parsing uploaded transcript PDFs to extract course data (codes, names, grades, credits) Generating personalized transfer advice based on the student's progress Data Pipeline: We built an indexer that processes 1,939 articulation agreement JSON files from ASSIST.org These are stored in SQLite for fast searching by university/major combinations The raw JSON files contain the full articulation mappings with course equivalencies Matching Algorithm: The core comparison logic handles: Course code normalization (e.g., "MATH 150" vs "MATH150") University name standardization (e.g., "UCLA" → "University of California, Los Angeles") Requirement group logic (handling "complete all" vs "select N from list" requirements) Section-level rules for complex prerequisite patterns

Challenges we ran into

Parsing ASSIST.org's data structure – The articulation agreement JSON files have deeply nested structures with courses embedded in sections, groups, and requirement blocks. Understanding how to extract the actual course mappings took significant reverse-engineering. Handling "Select N from list" requirements – Not all requirements are "complete all courses." Many say things like "Select 2 from the following 5 options." Building logic to properly calculate progress for these mixed requirement types was tricky. Transcript variability – Every community college formats their transcripts differently. Getting consistent course extraction from diverse PDF formats required careful prompt engineering with the AI model. Port conflicts on macOS – A fun debugging moment: Port 5000 is used by AirPlay Receiver on modern Macs, which broke our Flask server until we switched to port 5001!

Accomplishments that we're proud of

Real data, real impact – We're working with actual articulation agreements covering all 9 UC campuses and hundreds of majors Intelligent matching – Our algorithm correctly handles complex requirement structures, not just simple course lookups Clean UX – The interface is intuitive enough that any student can use it without instructions AI-powered insights – The recommendation system provides actionable next steps, not just raw data Speed – What used to take hours of manual research now takes seconds

What we learned

Domain complexity matters – The California transfer system has more edge cases than we initially imagined. "Select 2 courses from Group A OR 1 course from Group B plus 1 from Group C" is a real requirement pattern AI prompt engineering is an art – Getting Gemini to consistently return structured JSON from messy PDFs required iteration Data quality > quantity – Having 1,939 agreement files is useless if you can't parse them correctly Build for real users – Every design decision was guided by "Would this actually help a stressed student at 2 AM trying to figure out their classes?"

What's next for assist+

Short term: Support for more majors and course options in the dropdown Better handling of in-progress courses vs. completed courses GPA calculation and TAG (Transfer Admission Guarantee) eligibility checking Medium term: Semester planning tool – "Here's the optimal course schedule for your remaining requirements" Integration with community college course catalogs to show class availability Support for CSU transfers (not just UC) Long term: Mobile app for on-the-go planning Counselor dashboard so advisors can track multiple students Expand beyond California to other state transfer systems Our vision: Every transfer student should walk into their counselor's office informed, not anxious. assist+ is step one toward making that a reality.

Share this project:

Updates