AK

Amit kumar

AI Engineer

Gautam Buddha Universityfull_time, internship, freelance
Open to roles
PythonMachine LearningpandasSQL

EXPIRY ALERT ' 26

1. Clean Separation of Concerns Each branch corresponds to a single feature. The code is well-organized with lib/ for utilities, components/ for shared UI, app/ for pages and API routes. Easy to navigate and maintain. 2. Expiry Tracking Logic is Correct Status is computed dynamically (not stored), so it always reflects today's date. The 30-day threshold for "expiring soon" is a sensible default. Urgency sorting puts the most critical items first. 3. API Returns Computed Status Server-Side The records API enriches each record with status and daysUntilExpiry and sorts by urgency before sending to the frontend. This means the frontend just renders what it receives - no duplication of business logic. 4. Dashboard is Purposeful The dashboard does exactly what the problem statement asks: one screen showing what is active, what is expiring soon, what has expired, and what needs immediate action. The 4 summary cards give instant visibility. 5. Good Grouped View on Records Page Expired, Expiring Soon, and Active records are displayed in separate sections with color-coded headers. Within 7 days gets red urgency indicator, within 14 days gets yellow. This makes it easy to spot what needs attention. 6. Consistent Header with Global Search All pages share the same header with nav and search bar. The search page supports both text search and category/status filtering. 7. Clean UI, No Visual Noise Tailwind CSS is used consistently. Colors follow a logical green/yellow/red pattern. No emojis, no em dashes, no distractions. 8. Seed Data is Thoughtful 15 records across 8 categories with different expiry states (expired, soon, active) means the reviewer can immediately see all features working without creating data manually.

github.comView
Live Digital Queue System: Replacing Paper Slips with Instant Socket Sync

Live Digital Queue System: Replacing Paper Slips with Instant Socket Sync

Over 76% of India's 1.5 million neighborhood clinics rely on paper token slips and receptionist shouting. Patients wait 2-3 hours in crowded waiting rooms with zero visibility into their actual wait times. Receptionists manage queues entirely from memory during rush hours, leading to scheduling errors, administrative fatigue, and constant interruptions from patients asking, "When is my turn?" Process I designed and built a live sync client-server prototype using Node.js/Express, Socket.IO, and MongoDB. First, I structured the database schemas to support real-time token transitions. Second, I built a lightweight REST API for data mutations and integrated Socket.IO to broadcast state updates in under 500ms. Initially, I designed a dark cyberpunk theme but realized it lacked clinical credibility. I iterated to a clean, high-contrast light theme(utilizing Inter typography and soft slate backgrounds) optimized for receptionist monitors and TV waiting displays. Finally, I built dynamic wait time calculations using recent consultation averages. Results • Receptionist check-in times reduced to under 10 seconds using auto-focus forms and click-to-fill visit tags. • 100% accurate, zero-refresh live queue synchronization across all client screens in under 500ms. • Estimated wait times switch from a static 7-minute baseline to a dynamic, data-driven calculation after just 2 completed consultations. • Resilient fallback storage guarantees zero-configuration setup during evaluation. Reflection If I had more time, I would decouple the static frontend assets from the Express backend and host them on a CDN like Vercel to optimize asset loading. I would also add atomic MongoDB operations (findOneAndUpdate with transactional locks) to prevent race conditions in multi-doctor clinics where several receptionists might call patients simultaneously. Lastly, I would integrate SMS/WhatsApp notifications to alert patients when they are next in line.

1 media fileView
Student Expense Radar: Offline CLI Expense Tracker

Student Expense Radar: Offline CLI Expense Tracker

College students often make many small daily payments through UPI, canteen bills, mobile recharges, travel, and books, but they rarely track where the money goes. Most expense apps need internet, accounts, or a full mobile app, which is too heavy for a quick student use case. I built a simple Python command-line tracker that works offline, opens instantly, includes 3 sample expenses for reviewers, and helps students see total spending, highest-spend category, and budget warnings. Process I started by breaking the brief into must-have actions: add expense, view expenses, calculate total spend, identify the highest-spend category, and support a monthly budget warning. I kept the interface menu-based so beginners can use it without remembering commands. The first version stored expenses only in a list, but I realized that every new terminal session would lose the student’s history. I then improved it by saving expenses and budget data to a local JSON file using only the Python standard library. I also added a delete option because students may enter the wrong amount or category. Finally, I tested the app through multiple terminal runs, refreshed the sample output, and added a README plus screenshot for submission. Results The final project is a working offline Python CLI with 8 menu options, 3 preloaded sample expenses, local JSON persistence, category-wise spending insight, total spend calculation, delete support, and monthly budget warnings. It uses only the Python standard library, needs no database, no internet, and no setup beyond running the .py file. In testing, an expense added in one session appeared correctly in a later session, and deleting it updated the saved data. Reflection With more time, I would add monthly filtering so students can compare spending month by month instead of seeing only the full history. I would also add CSV export for sharing or analyzing expenses in Excel, and maybe a simple search/filter by category. Another improvement would be recurring expenses for recharges or subscriptions. I kept the current version focused on reliability and hackathon constraints, but the next version could become a fuller student money dashboard.

1 media fileView

This is Amit’s work on Wooble.

Build a profile that shows what you can do — and share it anywhere.

Build yours