name:        Vinit Girdhar
location:    Mumbai, India
role:        Final-year B.Tech IT @ St. Francis Institute of Technology (2027)
focus:       Machine Learning, Agentic AI, Full-Stack Engineering
currently:   Building LLM-powered agents and end-to-end ML systems
highlights:  SIH 2025 Grand Finalist | 4x Hackathon Winner | 20+ GitHub Stars

• I build end-to-end ML systems: demand forecasting, multilingual NLP, computer vision, OCR, and LLM-powered agents for civic, healthcare, and sales domains
• Smart India Hackathon 2025 Grand Finalist with Nirogya, plus 4 hackathon wins and 2 grand finalist finishes at national-level hackathons
• Deep into Agentic AI: LangGraph, Model Context Protocol (MCP), and multi-agent pipelines
• I love shipping products that create measurable real-world impact, especially for India-scale problems
• Reach me at girdharvinit786@gmail.com

Languages

Machine Learning and AI

Web and Backend

Cloud, Tools and DevOps

Predictive civic resource platform for Mumbai that forecasts urban demand 48 hours in advance and dispatches water tankers and garbage trucks across 65+ zones, powered by 51,440+ real municipal complaint records and a 25+ endpoint Flask API.

• Multilingual NLU pipeline in Hindi, English, and Marathi that ingests complaints from WhatsApp, voice calls, and web chat
• Equity engine that amplifies under-reporting low-income wards so silent communities are served proportionally
• Closed-loop ML feedback system that fires automatic model retraining alerts when prediction error exceeds 25%
• Addresses the Rs 1,500+ crore Indian municipalities spend annually on reactive, complaint-driven resource allocation

Tech: Python Flask Next.js Gemini 2.5 Pro Sarvam AI n8n SQLite

Repository | Live Demo

Smart India Hackathon 2025 Grand Finalist

Disease surveillance platform for rural and tribal communities that monitors waterborne diseases through real-time outbreak tracking, water quality monitoring, and community reporting.

• Multilingual, mobile-first reporting built for low-connectivity communities to flag symptoms early
• Enables health officials to detect and contain outbreaks before they spread
• Selected as a Grand Finalist at Smart India Hackathon 2025, India's largest national innovation competition

Tech: Machine Learning Real-Time Outbreak Tracking Multilingual Support Mobile-First

AI-based demand forecasting platform that predicts taxi demand 1 to 24 hours in advance using historical trip patterns, weather, public holidays, and event data.

• Processed and analyzed 100,000+ trip records to identify spatial and temporal demand trends across multiple city zones, integrating 5+ external data sources
• Demand heatmaps and hourly forecasts guide drivers to high-demand locations before demand peaks
• Reduces driver idle time by 20 to 35% and increases estimated daily driver revenue by 15 to 25%

Tech: Machine Learning Time Series Forecasting Geospatial Analytics Python

Repository

Full-stack AI sales intelligence platform where a conversational voice agent autonomously makes initial qualification calls and LLaMA 3 scores every conversation from 0 to 10 on the BANT framework (Budget, Authority, Need, Timeline).

• Targets the 60 to 70% of rep time typically lost on leads that never convert
• Auto-surfaces only high-intent prospects so sales teams talk exclusively to people ready to buy

Tech: ElevenLabs Conversational AI Groq LLaMA 3 Full-Stack

OCR and AI pipeline that automatically extracts transaction details from receipts and bills, categorizes expenses, and surfaces actionable spending insights.

• Eliminates manual data entry from personal finance tracking
• AI-driven expense categorization enabling smarter money decisions

Tech: OCR AI Expense Categorization Python

Repository

E-commerce platform that directly connects local street food vendors, such as vada pav and bhel puri sellers who lack the volume for large-scale orders, with wholesalers, giving micro-vendors access to wholesale supply chains without middlemen.

• End-to-end commerce flows covering vendor discovery, cart and checkout, and order tracking
• Wholesaler product management, profile management, and analytics for both sides of the marketplace

Tech: Flask Python Full-Stack E-Commerce

Repository

Offline-capable Chrome extension that intercepts uploads of 5 file types (.pdf .docx .xlsx .csv .md) across 4 major AI platforms (Claude.ai, ChatGPT, Gemini, Perplexity) and converts them locally in-browser into highly optimized Markdown before upload.

• Massive context window savings, lower API costs, and faster AI reasoning
• Removes platform-specific file size and file type friction
• All processing happens locally, with zero data leaving the browser

Tech: JavaScript Browser Extension Local File Processing