Suravi Sehgal

Aegis- stimulation of global trade

Geopolitical Energy & Supply Chain Risk Simulator Next.js | React | TypeScript | Tailwind CSS | Geospatial Visualization - Built a scenario-based decision-support prototype to simulate the impact of geopolitical disruptions such as blocked shipping routes, chokepoint closures, and oil supply disruptions. - Developed a custom scenario engine to model the chain from route disruption → supply loss/rerouting → fuel & oil price impact → logistics/business disruption → economic risk. - Designed interactive geospatial visualizations to display shipping routes, affected regions, exposure, and risk zones. - Added impact explanations and custom decision logic to help users understand the “why” behind the numbers and compare different scenarios. - Built an interactive dashboard using React + TypeScript, with responsive UI designed using Tailwind CSS. - Developed as a prototype using predefined assumptions and illustrative data, demonstrating how geopolitical and energy risks can be visualized for better decision-making.

Mira

🚀 MIRA — Intelligent Personal AI Workspace MIRA is a project built around a simple idea: what if your AI could actually help you get work done, instead of just answering questions? It brings everyday digital tasks into one intelligent workspace, helping users plan tasks, understand emails and documents, remember important context, manage schedules, and automate workflows. MIRA can take a request, understand what needs to be done, break it into smaller steps, use the relevant information, and assist with completing the task — while keeping the user in control through human approval when needed. 🤖✨ Built around: AI Agents • LLMs • Memory • Context Awareness • Workflow Automation • Email & Document Intelligence • Calendar Integration • Human-in-the-Loop AI GitHub: https://github.com/suravisehgal/mira-ai_iworkspace

PlanAgent- SAAS Landing Page

Built a modern, responsive SaaS landing page for PlanAgent, an AI-powered productivity assistant designed to help students, freelancers, and business owners organize tasks, plan schedules, and improve productivity. What I built: 1. Designed a clean, mobile-first landing page with responsive layouts across desktop and mobile. 2. Created a structured features section to communicate the product value clearly. 3. Built a pricing section, FAQ accordion, and animated testimonial slider to improve user engagement. 4. Added smooth scroll and fade-in animations for a polished browsing experience. 5. Implemented SEO basics including sitemap.xml, robots.txt, and a custom 404 page. 6. Integrated Google Analytics to track user behavior and page engagement. Github repo: https://lnkd.in/gtPR-iGc

Skincare Flutter Mobile App

Built a skincare e-commerce mobile application designed to help users browse products, view details, and manage purchases through a clean and intuitive shopping experience. What I built: 1. Developed a responsive product catalog with structured categories and product detail views. 2. Designed a smooth shopping flow with cart management and user-friendly navigation. 3. Implemented a modern UI focused on clarity, usability, and a premium shopping experience. 4. Connected the frontend to a backend database for product data management and dynamic content handling. 5. Optimized the app for mobile usability and smooth interaction across screens.

Timetable Generator – Python

Problem: Manual timetable planning is slow, error-prone, and hard to adapt when teachers are absent or rooms are unavailable. Solution: Building a school scheduling tool that centralizes teacher, room, and class data and helps automate timetable generation. Key Feature: 1. Building a Python-based scheduling system to simplify school timetable planning. 2. Centralizing teacher, room, and class data into a structured database for easier management. 3. Adding logic for room capacity, class size, teacher availability, and substitutions. 4. Designing features for leave management and dynamic timetable adjustments. 5. Working toward reducing manual scheduling effort and improving planning accuracy. Github Repository: 🔗 https://lnkd.in/gguy2ycP

SafeStep

GDG TechSprint 2026 Winner | Team of 3 Problem: Women navigating unfamiliar areas face safety risks, especially in poorly lit or isolated locations. Existing navigation apps optimize for speed, not safety. Solution: Built SafeStep, an AI-powered navigation platform that analyzes walking routes in real-time and generates a Safety Score based on lighting, foot traffic, and proximity to emergency services. Users can trigger one-tap SOS alerts to family and police with live location sharing. Key Contributions: 1.Integrated Gemini 1.5 Flash API for real-time route safety analysis, processing location data and generating safety scores for multiple route options 2. Built Guardian Network feature enabling users to add trusted contacts who receive live location updates during travel 3. Implemented one-tap SOS alert system with Firebase Cloud Messaging for instant notifications to family and local police 4. Developed route comparison UI showing safety scores alongside travel time, allowing users to make informed decisions 5. Deployed backend on Google Cloud Platform with Firebase for real-time data synchronization Impact: 🥇 1st Place Winner at GDG TechSprint 2026 (organized by GDG on Campus IFHE, Hyderabad) 🔗 GitHub: https://lnkd.in/gypQQYRC 🎥 Demo: https://lnkd.in/g97J8f6B 🌐 MVP: https://lnkd.in/gnADzXUn

Aqualens

Smart India Hackathon 2025 Finalist | Team OCEAN SIGHT Problem: Marine researchers struggle to collect data on rare species, while divers and ocean enthusiasts lack tools to identify and learn about marine life they encounter. Solution: Built AquaLens, a marine conservation platform that allows users to upload photos of marine species and instantly receive identification, habitat information, and geographical distribution data. Unidentified species are stored in a crowdsourced database accessible to researchers. Key Contributions: 1. Designed image recognition system using [specify: TensorFlow, pre-trained models, or custom CNN] to identify marine species from user-uploaded photos 2. Built crowdsourced database enabling users to contribute data on rare or unidentified species, which is made accessible to marine researchers 3. Implemented gamification system awarding credits to users for contributions, especially for rare or newly documented species 4. Created detailed species information pages with habitat maps, conservation status, and historical data 5. Developed researcher dashboard for accessing crowdsourced data, filtering by species, location, and time period Impact: 🥈 2nd Position in Smart India Hackathon 2025 (Internal Round)

Side_scan_Sonar

An AI-powered system that analyzes side-scan sonar images to detect and classify underwater objects such as marine debris, rocks, and other seabed features. Its uniqueness lies in automating sonar image interpretation, reducing reliance on manual inspection, and distinguishing between different underwater objects to support faster, more efficient seabed exploration, marine debris identification, and underwater surveying.

Polaris

🚢❄️*POLARIS (Polar Intelligence, Risk Assessment & Intelligent Navigation System)* is an AI-enabled decision-support platform designed to make Antarctic research-vessel navigation more *predictive, risk-aware, and resource-efficient* What we’re building: • AI-based sea-ice concentration forecasting using historical & current environmental data • Satellite-based iceberg detection and trajectory prediction • Dynamic navigation risk assessment using ice, iceberg, weather, ocean currents & vessel parameters • A*-based multi-objective route optimization considering distance, travel time, fuel consumption and environmental hazards • Dynamic rerouting as new environmental observations become available • Interactive Antarctic dashboard for visualizing ice conditions, iceberg trajectories, risk zones, alerts and alternative routes Our goal is to transform diverse Antarctic environmental data into *actionable navigation intelligence*, while keeping qualified navigation personnel in control.

Stock_alerts

Introducing Stock Alerts: ✨ Key Features: 🔹 Smart Stock Filtering: Classify and filter stocks based on RSI levels, stock prices, market capitalization, and market performance. 🔹 Personalized Watchlist: Choose the companies you’re interested in and track them through a customized dashboard. 🔹 Buy/Sell Alerts: Receive pop-up notifications when stocks meet relevant buy or sell conditions, reducing the need to check prices manually. 🔹 Individual Profit & Loss Tracking: Monitor individual stock performance and understand price movements. 🔹 Risk Analysis: Assess the potential risk associated with a stock before making an investment decision. 🔹 Market Trend Analysis: Analyze market movements and identify potential buying and selling opportunities using technical indicators and predictive insights. 💡 The core idea? Less time searching. More personalized insights. Better-informed decisions. This project reinforced my belief that technology should not just provide more data; it should help people understand the data that matters to them.

ExamShield

Description: ExamShield is an automated, AI-driven online proctoring and assessment integrity platform designed to prevent academic malpractice during remote examinations. In virtual testing environments, manual invigilation struggles to catch subtle cheating behaviors. ExamShield bridges this gap by continuously monitoring the test environment in real time—tracking eye gaze shifts, flagging head-pose anomalies, detecting the presence of multiple faces or candidate absence, and enforcing browser lockdown against unauthorized tab and application switching. It delivers a tamper-resistant, secure evaluation pipeline that ensures fair grading and transparent incident reporting for educational institutions and certification bodies. Key Capabilities: Real-time gaze & facial tracking, browser lockdown integrity, multi-person detection, automated violation logging, and low-latency anomaly alerts. Tech Stack: Python, Computer Vision (OpenCV), Image Processing, Machine Learning.

vyra_false.pods

vyra_false.pods is a specialized computer vision and peripheral auditing tool developed to eliminate one of the hardest-to-catch cheating loopholes in remote assessments and technical interviews: concealed wireless earbuds and spoofed audio devices. While conventional proctoring tools detect browser switching or camera obstruction, candidates frequently bypass security using hidden micro-earpieces, AirPods concealed under hair, or virtual audio driver redirection ("false pods"). vyra_false.pods employs facial landmark detection and localized ear-region visual scanning to identify unauthorized wearable audio hardware, while auditing connected device feeds to flag covert external assistance in real time. Key Capabilities: Concealed earbud/in-ear hardware recognition, ear-region spatial landmark tracking, rogue audio device routing detection, and instant invigilator alert triggers. Tech Stack: Python, Computer Vision, Landmark Detection, Peripheral Device Auditing.

Infosys Springboard