UniFlow — AI University Companion

UniFlow is a chat-first AI companion built for the 1.3 million Malaysian university students drowning in scattered admin — assignments on Spectrum, deadlines on WhatsApp, budgets in their head, scholarships nobody tells them about, and internships buried across the web. You text UniFlow like a friend. "Calc quiz Thursday 2pm, spent RM15 on lunch, find me CS scholarships" — and your schedule, budget, assignments, and opportunities update instantly. The product is built on three production-grade layers: (1) Bright Data SERP API powers our Scholarships and Internships tabs with live, real-time web intelligence. Scholarships runs 4 parallel queries against Malaysian sources (JPA, MARA, Yayasan, private foundations). Internships runs 8 parallel queries with tier-1 company boosting for Grab, Shopee, TikTok, Maybank, Petronas, AirAsia, and CIMB — deduplicated and diversity-enforced. (2) Groq llama-3.3-70B with function calling parses natural language into structured events under 500 milliseconds and extracts company, role, eligibility, and deadlines from raw SERP results. (3) A Chrome Extension (Manifest V3) reads Universiti Malaya's Spectrum Moodle DOM, imports courses, assignments, and study materials — auto-classified by file type using icon and filename analysis — with zero credential handling. Built end-to-end in vanilla HTML, Tailwind CSS, FastAPI, and Firebase Auth + Firestore for real-time multi-device sync. UniFlow spans Track 1 (GTM Intelligence — surfacing scholarships and opportunities) and Track 2 (Finance & Market Intelligence — internship discovery and budget tracking). Built by two Year 1 Computer Science students at Universiti Malaya who needed it themselves. This isn't a hackathon project. It's the startup we ship Monday.