I'm Gurpreet Singh, an AI Automation Engineer based in Raipur, India — I design and build production-ready AI systems that eliminate manual work: voice agents that handle inbound calls, lead generation pipelines that extract 500+ verified contacts per session, and intelligent automation workflows that run 24/7 without human input.
Founder of AutoFlow Automation Systems (Dec 2025–Present), where I ship AI voice receptionists, multi-channel automation pipelines, and workflow orchestration systems for businesses. Simultaneously working as a Data Scientist at Sabudh Foundation, applying Python, ML, and data pipelines to real-world problems.
My stack: n8n, Google Gemini API, Voice AI, Python, REST APIs, WhatsApp/Gmail integrations — everything wired together into systems that businesses can actually deploy and rely on.
Milestone conversational AI agents — live & interactive
Conversational AI that recommends movies by your mood, captures booking details through natural dialogue, auto-allocates seats, and confirms via WhatsApp & Gmail. Zero human in the loop.
24/7 AI receptionist for dental clinics — handles patient queries, books appointments through natural conversation, and sends WhatsApp confirmations. No human required.
A fully automated AI cinema booking agent. Rhea understands your mood, recommends matching movies, captures your booking preferences through natural conversation, allocates seats in real time, and sends dual confirmation via WhatsApp and Gmail — all without any human involvement. Built on n8n, Google Gemini, and WhatsApp Business API. Zero manual steps across the entire booking flow.
A 24/7 AI voice receptionist purpose-built for dental clinics. Sarah handles patient queries round the clock, books appointments through natural voice conversation, verifies time slots, and sends instant WhatsApp confirmations — completely replacing front-desk staff after hours. SaaS trial onboarding logic included for demos. Simulation showed 70% fewer missed patient calls with zero additional staffing cost.
A scalable B2B lead generation pipeline that automatically extracts business names, emails, phone numbers, websites, and social URLs from Google Maps using n8n and the Maps API. Data is cleaned, deduplicated, and pushed to Google Sheets in real time. Generates 500+ verified, outreach-ready leads per session and cuts manual prospecting time by 80%. Adaptable to any niche — clinics, agencies, retail, real estate.
A production-grade RAG chatbot that lets you query any custom knowledge base in plain language. Documents are chunked, embedded, and stored in a vector database. Each query retrieves the most semantically relevant chunks and injects them into Google Gemini as context — delivering accurate, document-grounded answers with full source traceability. Ideal for internal wikis, product docs, legal FAQs, and support automation.
An end-to-end ML pipeline built during the Sabudh Foundation internship. Performed thorough EDA on the Titanic dataset, engineered features including family size groupings, passenger title extraction, and age-bin encoding. Trained and benchmarked Random Forest, Gradient Boosting, and XGBoost classifiers using cross-validation and hyperparameter tuning — demonstrating a complete workflow from raw data to production-ready model.
Automated Instagram scraper that extracts business profiles, contact info, follower counts, engagement rates, and niche keywords — structured into Google Sheets. Filters by location, category, and follower range to deliver high-quality, targeted leads for outreach campaigns.
Extracts 500+ verified B2B leads per run from Google Maps — business names, emails, phone numbers, websites, and social profiles — delivered directly into Google Sheets in minutes. Supports any city, category, or keyword. Built with Python and n8n for fully automated, repeatable prospecting.
A fully automated cold outreach system. Scrapes targeted leads, uses Gemini AI to craft personalized emails per prospect, sends via Gmail API, and tracks opens, clicks, and replies — all without manual effort. Includes follow-up scheduling and reply detection for a complete sales pipeline.
Full-stack B2B intelligence pipeline: Google Maps extraction → AI-powered lead enrichment with company summaries and pain-point insights → structured Google Sheets output. Delivers 500+ enriched, context-rich leads per session with 80% less time than manual prospecting. Ready for immediate outreach.
AI-powered caption engine for social media. Analyzes image content and context using Gemini Vision, generates tailored captions for Instagram, Twitter, and LinkedIn — each with platform-optimized tone, length, and hashtag sets. Batch-processes multiple posts and delivers outputs to Sheets or Notion.
Analyzes trending YouTube topics, competitor channels, and keyword gaps to generate ranked video ideas — complete with attention-grabbing titles, proven hook formulas, and full script outlines. Backed by Gemini AI and YouTube Data API. Outputs a weekly content calendar ready for production.
Scrapes top-ranked HackerNews stories daily, generates concise AI voiceover scripts using Gemini, creates professional talking-head avatar videos with AI video cloning tools, and auto-publishes to YouTube with SEO-optimized titles and descriptions — a complete content machine running on autopilot.
Complete YouTube Shorts automation pipeline: idea input → AI scriptwriting → Veo AI video generation → audio sync → thumbnail creation → YouTube upload → performance tracking in Google Sheets. Turns a single idea into a published Short in minutes with zero manual production work.
AI-powered wedding invitation generator that creates fully personalized, culturally accurate cards based on religion, language, region, and couple preferences. Gemini AI writes the copy; image generation models produce print-ready designs. Supports Hindu, Muslim, Sikh, and Christian formats in multiple languages.
A Telegram-to-YouTube automation bot. Monitors a Telegram channel or group for video uploads, downloads them automatically, uses Gemini AI to generate SEO-optimized titles, descriptions, and tags, then uploads directly to YouTube via the API with full metadata. Ideal for content teams and creators repurposing short-form video.
A conversational AI that acts as your personal scheduling assistant. Books, reschedules, and cancels appointments via natural voice or text dialogue, checks calendar availability in real time, sends WhatsApp or email confirmations, and fires automated reminders before each appointment. Zero manual calendar management required.
An autonomous job application agent that scrapes targeted job listings from multiple platforms, uses Gemini AI to tailor your resume and write a personalized cover letter for each role, and submits applications automatically. Tracks all applications in a Google Sheet with status updates — scaling your job search without the grind.
A context-aware personal AI assistant with persistent memory and multi-tool access. Remembers past conversations, manages tasks and to-dos, answers questions using RAG over your personal knowledge base, and adapts to your preferences over time. Integrates with Google Calendar, Gmail, and Notion for a unified productivity layer.
A fully automated spiritual content engine. Generates daily quotes sourced from sacred texts and wisdom traditions using Gemini AI, pairs them with AI-generated visuals matching the theme, and auto-schedules posts across Instagram, Facebook, and Twitter. Runs completely hands-free on a daily cron schedule.
Fetches breaking AI and tech news from 10+ sources including RSS feeds, Reddit, and HackerNews, filters by relevance, uses Gemini to summarize and categorize each story, and compiles a clean daily digest delivered via email or Telegram. Keeps you informed in 5 minutes without information overload.
A production-ready RAG chatbot that queries any custom knowledge base using natural language. Documents are chunked, embedded, and indexed in a vector store. Queries retrieve semantically matched chunks that are injected into Gemini as context — producing accurate, citation-backed answers. Deployable as a web widget, Telegram bot, or API endpoint.
A multi-agent advisory system for technology decision-making. A routing agent classifies each query and dispatches to specialized sub-agents: a SQL agent for structured trend data, a RAG agent for research papers and reports, and a synthesis agent for final recommendations. Helps engineering leaders make data-backed technology adoption decisions.
Full end-to-end ML project from the Sabudh Foundation internship. Deep EDA uncovered survival patterns by class, gender, and age. Feature engineering included title extraction, family groupings, and cabin-deck encoding. Benchmarked Random Forest, Gradient Boosting, and XGBoost with GridSearchCV tuning — delivering a reliable survival prediction model from raw, messy historical data.
Autonomous multi-agent framework where specialized AI agents collaborate — one researches, one writes, one reviews — completing complex tasks with zero human intervention.
Your all-in-one AI companion — writes emails, reviews resumes, generates content, updates Sheets, books appointments, and handles 10+ daily workflows autonomously.
Have a process that needs automation? An idea that needs an AI agent? Open for freelance work and collaborations.