Software Engineer

github

Ayush Jain  • Peerlist

"Empowering Fitness Journeys with AI-Driven Plans and Expert Guidance, Anytime, Anywhere"

Activ-AI-te

Asp-AI-re

"Empowering Developers to Collaborate, Communicate, and Create Seamlessly"

DevXP

"Transforming Document Management with AI-Driven Summaries, Insights and Smart Scanning" • Built DocuMate, an AI-powered PDF analysis platform enabling users to chat with documents—reduced manual document review time by up to 80% and increased information retrieval speed by 3x using real-time Q&A, semantic search (TF-IDF + cosine similarity), and Groq-powered insights. • Developed a scalable FastAPI-React system integrating LangChain and Groq API to process large PDF files (up to 500+ pages) with chunked vectorization, achieving 95% accuracy in contextual answer matching and reducing average query response time to less than 2 seconds. • Designed a fully responsive, mobile-first UI with toast notifications and markdown-rendered AI responses—enhancing user satisfaction, leading to a 40% increase in engagement time and seamless document interaction across all screen sizes.

DocuMate

"Connecting Farmers & Cultivating Prosperity through AI-driven Insights, and Virtual Support" • Developed FarmEzy, an AI-powered full-stack MERN application empowering farmers with personalized agricultural insights—integrated a multilingual RAG-based chatbot (AgriAssistant) built on Gemma 3B:4b, enabling 90%+ accurate voice/text interactions across multiple regional languages. • Implemented AI-driven crop disease detection using CNN models served via Flask APIs—achieved 92% accuracy in plant disease prediction from leaf images, accelerating diagnosis and reducing crop loss risk. • Integrated a Smart Irrigation System and real-time Weather API to provide location-based irrigation insights, improving irrigation efficiency by 30%, enhancing resource consumption. • Built a competitive marketplace module, increasing crop value discovery potential by 25–40% through transparent price comparison.

FarmEzy

"Empowering Financial Wellness through Smart Insights, Seamless Tracking, and AI-Driven Guidance"

Finnest

GoKart

Legalyz

LogiX

Developed and optimised a personal portfolio website, hosted on GitHub,  secured by cloudflare, did some SEO with the help of Google Analytics.

Portfolio Website

• Designed a centralized platform for Gujarat’s innovation ecosystem, streamlining collaboration and data management among key stakeholders: startups, researchers, policymakers, investors, and IPR managers, resulting in a 30% improvement in operational efficiency. • Engineered role-specific dashboards and a real-time monitoring system for researchers, startups, and policy-makers, enhancing project visibility and reporting accuracy by over 25%. • Built modules for progress tracking, reward systems, and data-driven insights, achieving a 20% increase in transparency for project outcomes and resource allocation through real-time analytics. • Integrated a dynamic IPR management module that automated application workflows and status tracking, leading to a 25% reduction in patent filing timelines and improved compliance visibility.

RISE (Research, Innovation, Startups & Entrep

"Bridging Distances & Enhancing Care by Empowering Health through AI, IoT, and Virtual Connectivity" • Developed an AI-powered healthcare platform bridging the gap between patients & doctors via virtual consultations, achieving 70% reduction in travel barriers and delays. • Implemented secure, real-time appointment orchestration using Google Calendar and Meet/Zoom APIs, reducing scheduling latency by 80% and ensuring high availability of consultation slots. • Integrated disease-specific predictive analytics models based on CNN and NLP for early detection of chronic illnesses (e.g., cardiovascular diseases, diabetes, brain hemorrhage, skin diseases, etc.), delivering over 90% predictive accuracy and enabling timely clinical interventions. • Designed Dr. MediBot, a multimodal Retrieval-Augmented Generation (RAG) conversational agent built on FAISS-based vector search, capable of processing text, speech, and PDF queries, resulting in a 40% improvement in autonomous user query resolution. • Engineered four AI-driven features using MedLLaMA: prescription summarization with 99% accuracy, drug recommendation, personalized health tips, and symptom-based disease prediction with 92% precision, collectively improving diagnostic efficiency and patient engagement by 45%.