Full-stack developer & IT undergrad crafting platforms that make AI, IoT, and intelligent systems accessible — from misinformation detection to agricultural tools to environmental monitoring.
Projects
Each project tackles a real problem — from countering misinformation to empowering farmers to tracking environmental crises.
A production-grade testing platform that supports both objective (MCQs) and subjective evaluations with AI-assisted grading. Enables administrators and moderators to create, manage, and evaluate structured assessments with a clean, distraction-free test-taking UI.
Built with a focus on real-world system design, including server-synced timers, fault-tolerant auto-save mechanisms (every 30s + navigation), cursor-based pagination, and middleware-based role enforcement without extra DB calls.
Full-stack platform with autonomous multi-agent AI architecture (Eyes, Brain, Knowledge Core & Tongue) for real-time misinformation detection. Multi-agent debate system, government dashboard, WhatsApp bot, knowledge graph, 200+ language support.
Helps citizens, journalists, and institutions verify claims within minutes. Addresses fake news spreading during elections and health emergencies. Research paper authored on its architecture.
Real-time platform with ESP32 + Raspberry Pi sensor network (MQ135, LDR, IR, pH), 3D Earth model, deforestation heatmaps, VAPI voice agent, and crowdfunding for marine/forest conservation.
Tracks environmental crises in real-time — oil spills, deforestation, plastic pollution — and connects NGOs, government, and donors to fund conservation. 94% project success rate.
Multilingual platform for India's farmers — AI crop disease diagnosis via Google Gemini, real-time weather alerts via Twilio, intelligent crop planning, government schemes directory, and community features.
Bridges the gap between farmers and technology. Helps diagnose crop issues, boost yields, access government schemes, and get expert guidance for rural India.
Three-app platform — Customer (groceries + medicines), Shopkeeper (AI invoice scanning via Donut model, barcode sync via zxing.js), Delivery. Auto-extracts stock data from invoice photos, eliminating manual entry entirely.
Digitizes neighborhood kirana stores and pharmacies, enabling them to compete with large e-commerce platforms while serving local communities efficiently.
Combines smart wearable data (temperature, HRV, sleep, activity) with predictive AI giving 3–5 day advance cycle alerts. Personalized PCOD/PCOS care kits. Targeting 355M+ menstruating women in India.
Replaces unreliable calendar tracking with body intelligence for women with irregular cycles. 70M+ Indian women have PCOS/PCOD with no predictive tool built for them.
Colortoons is the React Native color-by-number mobile game. ColorArt is the FastAPI backend powering it — MiniBatchKMeans quantization → OpenCV segmentation → Pole of Inaccessibility labeling → Chaikin/Catmull-Rom SVG smoothing.
Converts any image into an interactive SVG color-by-number map without heavy AI models. Fast, reliable, and low resource usage.
Skills
Technologies I work with — only what I've actually built things with.
Frontend
Mobile
Backend & Languages
Database & Cloud
AI / ML
IoT & Hardware
APIs & Integrations
DevOps
Education
B.Tech in Information Technology
Heritage Institute of Technology, Kolkata · 2024 – 2028
Higher Secondary (ISC – Science)
Sister Nivedita Institute, Howrah · 2023
High School (ICSE)
Sister Nivedita Institute, Howrah · 2021
Open Source
Live contribution data from my GitHub — @S0SP · 26 public repos
Contact
Have an opportunity, idea, or want to collaborate? Reach out.