Aman Kumar

We built this Farm-to-Fork delivery platform during a hackathon using only pure HTML, CSS, and JavaScript, along with Firebase for authentication, product storage, dashboards, and basic tracking. Since AI tools were not allowed in the hackathon, we focused completely on manual development, ensuring higher precision and fewer mistakes through clean, simple code. To achieve fast performance, we implemented a Single Page Application (SPA) structure, where all sections load from one page, enabling smooth navigation and efficient data handling without unnecessary reloads. One of the most impactful features we added was language switching — Hindi ↔ English, making the platform farmer-friendly and easy to use for local users. We kept the UI intentionally simple, clean, and distraction-free, avoiding fancy animations and effects. The goal was pure problem-solving, ensuring farmers can list products easily and restaurants can place quick orders without confusion. This project helped us understand real-world workflows, user-centered design, and building complete functional systems under time constraints.

🌐 HelpNet – A Crowdsourced Disaster Management System Built to Save Lives 🚨 HelpNet is a purpose-driven platform we built to empower communities during disasters and emergencies. Designed with accessibility, real-time response, and collaboration in mind, HelpNet bridges the gap between affected individuals, responders, and volunteers. 🛠️ What HelpNet Does: 📍 User-Powered Disaster Reporting – Individuals can report incidents with precise location data and optional media. 🧠 AI-Driven Verification – Uses machine learning to filter and validate incoming reports. 🗺️ Live Mapping Interface – Visualizes active disaster zones using Google Maps API. 🔔 Real-Time Alerts – Keeps users informed when they are near affected areas. 🙌 Volunteer & Relief Coordination – Connects NGOs, authorities, and volunteers to the right locations instantly. Our goal with HelpNet was to leverage technology for real-world humanitarian impact — transforming the way information flows during crises and enabling faster, smarter responses. 🏗️ Built With: JavaScript, Firebase, Google Maps API, HTML/CSS, and basic AI/ML models. This project has been a powerful learning experience in applying tech for social good, and we’re proud of what it represents. HelpNet continues to inspire our commitment to building solutions that matter. 📩 Always open to feedback, collaboration, or partnerships that can take HelpNet further.

It was Developed during my SDE internship, this tool extracts structured data from dynamically loaded websites without requiring code changes for different pages. It intelligently detects tables, lists, pagination, and infinite scrolling — making the scraping process fully automated and user-friendly. ✅ Key Features :- ✔ GUI-based (no coding needed) ✔ Auto-detects page structure ✔ Handles dynamic & lazy-loaded content ✔ Excel export with clean formatting ✔ Optional image download + verification ✔ Safe STOP with partial data saving ✔ Duplicate prevention via normalization ✔ Works in headless or visible mode 🛠 Tech Stack & Python Libraries :- - playwright - asyncio & threading - tkinter - pandas + openpyxl - httpx - logging 🖥️ How Users Get Results :- 1️⃣ Enter target webpage URL 2️⃣ Provide desired Excel filename 3️⃣ Configure options (images, headless mode, limits) 4️⃣ Click Start Scraping 5️⃣ Data is extracted & validated automatically 6️⃣ Results are saved directly to Excel ✅ If stopped, partial output is safely preserved 📂 Output Includes - auto-generated serial numbers - extracted text, links, and images - validation status (Valid / Broken / Missing) - multi-sheet support when multiple structures are found

🚀 Built a Typing Speed Test Web App using just HTML, CSS & JavaScript – no libraries, no backend. ✔ Real-time WPM & accuracy tracking ✔ Dynamic text passages ✔ Error highlighting ✔ Fully responsive UI A simple project that sharpened our core frontend skills and delivered a clean, fast typing experience!

Built a document Q&A system using LLMs, RAG, and vector embeddings with FAISS semantic search. Optimized retrieval pipelines with query caching, reducing document search time by 65%. Tech stack: Python, LangChain, FAISS, OpenAI API, Streamlit.

Built automated resume screening using NLP with spaCy, achieving 92% accuracy and TF-IDF ranking. Enabled batch processing with Excel exports, reducing manual screening by 70%. Tech stack: Python, spaCy, scikit-learn, Pandas, Streamlit.

Developed a Python scraper with automated validation and Excel export, improving reliability. Tech stack: Python, Selenium, BeautifulSoup, Tkinter, openpyxl.