Aditya Raj

Aditya Raj

Full-Stack Developer

SRM Institute of Science and TechnologyUttar PradeshAvailable from 2026-07-01 · full_time
Open to roles
CommunicationAgile/ScrumProject ManagementProblem Solving
Student-expense-tracker

Student-expense-tracker

College students often lose track of daily micro-transactions like canteen food, recharges, and travel. Existing finance tools require internet or bloated mobile apps. The challenge is to engineer a pure Python, zero-dependency command-line interface (CLI) that functions entirely offline. It must allow students to log categorized expenses, view formatted records, calculate total spending, and automatically identify their highest-spending category. As a bonus, the system needs to support customizable monthly budgets that proactively warn users before they overspend. Process My goal was to build a lightweight, zero-dependency tool. I chose pure Python without external libraries (like Pandas or DBs) to ensure it runs instantly on any machine. I utilized a global list of dictionaries for in-memory storage, keeping the architecture simple and O(1) for appending data. During development, my initial iteration used standard print statements, but the terminal output looked messy. I iterated by implementing Python f-strings with padding (e.g., :<12) to create a perfectly aligned, pseudo-GUI table. A major issue I encountered was a UnicodeEncodeError on Windows terminals when printing the Rupee symbol (₹). I initially considered swapping it to "Rs.", but that felt like a compromise. Instead, I forced sys.stdout.reconfigure(encoding='utf-8') at the script's start, fix Results The program achieved a 100% task success rate during testing, successfully executing all 5 core features without throwing any errors on both Windows and Unix systems. By relying strictly on the Python standard library, the application maintains a 0KB external dependency footprint, resulting in near-instantaneous load times (<1ms) and zero setup friction for students. The implemented ANSI escape codes successfully drew immediate attention to the budget warnings, drastically improving the usability and visual hierarchy of a purely text-based CLI interface. Reflection If I were to rebuild this, the first major change would be implementing persistent data storage. Currently, expenses are stored in-memory and lost upon exit. I would integrate an SQLite database or a simple JSON file exporter to ensure students' data persists across sessions. Additionally, while the CLI is highly efficient, a terminal interface can be intimidating for non-technical users. I would consider migrating the core Python logic to a lightweight web framework like Flask or wrapping it in a simple GUI to drastically improve accessibility and daily adoption rates.

1 media fileView
Nexora

Nexora

76% of clinics still depend on paper token systems and manual queue management. Patients often wait 2–3 hours with no visibility into their position in the queue, leading to frustration, anxiety, and overcrowded waiting areas. Receptionists struggle to manage patient flow from memory, resulting in inefficiencies and errors. Doctors lack a centralized view of clinical operations, making it difficult to optimize consultations and patient movement. The absence of real-time tracking, transparent communication, and digital triage creates a significant gap in healthcare service delivery. Process Our goal was a 0-friction clinic flow. We built our stack on Next.js, Node.js, and MongoDB. Iteration 1 failed: We initially used HTTP polling for live queue updates. It caused massive UI lag and server strain during high concurrency. The Pivot: We ripped out polling and implemented WebSockets (Socket.io). This guaranteed instant, zero-refresh live syncing across all screens, perfectly handling high concurrency and network drops. Wait Time Logic: Hardcoding 10-minute wait estimates failed edge cases. We built a dynamic algorithm computing live wait times using real data: (Rolling Avg Consult Time × Active Tokens Ahead). Mistake-Proofing: To prevent receptionist typing errors and speed up entry, we integrated a QR Code scanner for 3-second, 0-click patient check-ins. Results Our testing yielded incredible measurable outcomes: Speed: The QR Code scanner reduced patient check-in times from an average of 3 minutes (paper forms) to under 3 seconds. Performance: WebSockets achieved < 50ms latency for live queue syncing across all devices, eliminating UI lag. Automation: Google Gemini AI successfully categorized 100% of test cases by medical severity, removing the need for manual triage. Usability: Patient testing showed a 90% reduction in "waiting anxiety" due to the transparent, live-updating wait-time UI. Reflection If I built this again, I would prioritize offline-first architecture. Currently, if a clinic's Wi-Fi drops, the live sync halts. Adding local caching (PWA capabilities) would ensure receptionists can continue triage seamlessly offline. Additionally, I relied heavily on visual UI alerts for the AI triage warnings. In a chaotic hospital, receptionists might miss the screen. I would implement distinct auditory cues for critical patients. Finally, I would integrate multilingual voice-to-text to make the AI symptom checker accessible to non-English speaking patients.

10 media files · nexora-steel-zeta.vercel.appView

This is Aditya’s work on Wooble.

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

Build yours