Harish2859/hydro_drop
A lightweight, offline-first personal hydration tracker and interval reminder built with Flutter and Hive
"Success is not final, failure is not fatal: It is the courage to continue that counts." — Winston Churchill
A lightweight, offline-first personal hydration tracker and interval reminder built with Flutter and Hive
NumPy aware dynamic Python compiler using LLVM
14MB foundation model for tiny devices; phones, wearables, smart home, and robots.
A production-grade Legal Document Intelligence platform powered by Agentic RAG. Upload complex contracts (NDAs, Leases, SaaS agreements) to instantly detect "Red Flag" clauses, ask plain-English questions, and receive cited, verified answers anchored directly to the source text.
Full-stack Todo application built with Kotlin, featuring a Spring Boot REST API backed by PostgreSQL (Docker) and an Android Jetpack Compose client with Retrofit.
Lightweight local RAG implementation with pgvector and Gemma 4.
Stock Market AI Predictor is an advanced machine learning application that uses deep LSTM neural networks to predict stock prices with real-time accuracy. Features include interactive charts, technical analysis, bull/bear sentiment indicators, professional company branding, and a sleek dark-themed interface for comprehensive market insights.
Developed a Convolutional Neural Network (CNN) for automated brain tumor detection in MRI scans, achieving ~94% diagnostic accuracy and reducing false positive rates by ~12%. Optimized image preprocessing and data augmentation pipelines for complex MRI datasets, improving model generalization and reducing training convergence time by ~25%.
High-precision LLM for structured data extraction. Uses dual-phase SFT and DPO to transform unstructured text into validated JSON. Optimized for NVIDIA Blackwell (RTX 50-series) via Unsloth and QLoRA. Achieves 100% schema adherence with zero conversational hallucination. Ideal for high-reliability RAG and automated data pipelines.
A high-performance Neural Collaborative Filtering (NCF) recommendation system built with PyTorch, optimized for NVIDIA Blackwell (RTX 5050), and designed for AWS cloud deployment.
A sleek, high-performance Chrome Extension for LeetCode & CodeChef that uses Groq LPU inference to provide Socratic debugging and real-time memory visualization for new coders.
A secure, multi-role Agri-Fintech platform engineered with an Android Jetpack Compose frontend and a high-precision Spring Boot/PostgreSQL backend. Features milestone-based capital escrow routing, automated governance fee cuts, and real-time portfolio ROI reconciliation logic.
High-precision transactional backend core for the AgriInvest ecosystem built with Spring Boot and PostgreSQL. Implements secure multi-role JWT authentication, idempotent milestone fund release tracking, and automated platform fee distribution using BigDecimal math logic
This project is a beginner-friendly example of how a simple React application can be automated using a complete CI/CD pipeline. It shows how code pushed to GitHub can trigger Jenkins, build the app, package it into a Docker container, and serve it using Nginx.
A full-stack web application built as a first Docker integration project. It features a user profile page with editable fields, a Node.js HTTP server, MongoDB for persistent storage, and Mongo Express as a visual database admin UI — all wired together using Docker Compose
Real-time website uptime & performance monitor with concurrent polling, WebSocket live updates, and SSRF-protected URL checks. Built with React, Node.js, Express & PostgreSQL.
A LangGraph-based agent that analyzes Indian stocks (NSE/BSE) and produces a structured, evidence-backed briefing — fundamentals, sector comparison, recent news, and risk flags — with an independent verification layer that audits the agent's own claims against ground-truth data before anything is shown to the user.
A multi-tenant, enterprise-grade Kanban board built with the PERN stack, featuring role-based access control, real-time collaboration, and agentic AI.
The Cyber-Stock PC Inventory Engine is a high-performance, full-stack management system designed for custom PC builders and hardware retailers. Built with a focus on Data Integrity and Optimistic UI, this application moves beyond a simple list and functions as a smart rule-engine that understands hardware compatibility and power requirements.