I'm Prateek. I have 11+ years of experience, and I build AI systems that actually make it into production. That distinction matters. A lot of AI work looks great in a demo and falls apart the moment real users and messy real data hit it. Most of what I get hired for is either building something properly from scratch or fixing a system that worked in testing and stopped working in the wild. Here is the kind of work I take on. CHATBOTS AND ASSISTANTS THAT KNOW YOUR BUSINESS I build RAG systems that sit on top of your own documents, databases, and internal knowledge, so answers come from your data instead of the model guessing. Also, document processing, pulling structured information out of messy files, and semantic search across content that keyword search cannot handle. - OpenAI, Claude, Gemini, Llama, Mistral - LangChain, LlamaIndex - Pinecone and other vector databases AI AGENTS I build multi-agent systems with proper memory, planning, and tool use, where the agents actually complete a business process rather than just talking about it. This is most of my work right now and the part I enjoy most. - Multi-agent orchestration - Tool calling and function calling - Model Context Protocol - LangGraph, CrewAI AUTOMATION If repetitive manual work is eating your team's time, I automate it. Sometimes that is n8n or Zapier wired into your existing tools, sometimes a custom Python pipeline when the off-the-shelf route will not hold up. - Lead handling and support triage - Reporting and data entry - Content pipelines - n8n, Zapier, Make, custom Python PREDICTIVE MODELS If you have data and want it to predict something, I build the model. I recently built a system that matched warehouse labels out of unstructured text for a client in the US. - Forecasting and time series - Recommendation engines - Churn prediction and customer segmentation - Fraud and anomaly detection - Classification and entity matching - scikit-learn, XGBoost, PyTorch, TensorFlow COMPUTER VISION If your problem involves images or video, I handle that too. - Object detection and tracking - OCR and document parsing - Image classification - Facial recognition - Quality inspection and video analytics DEPLOYMENT Once it is built, I handle the production side properly. - APIs and containerisation - AWS and GCP hosting - Monitoring and retraining when the model drifts - FastAPI, Docker, MLOps HOW I WORK You get the architecture, cost, and timeline before you commit You see something working early, not after three silent weeks Everything ships as a working, deployed system with an API and an interface I stay reachable after delivery, because real usage always needs tuning I also take standalone web and mobile work: WordPress, Elementor, WooCommerce, Shopify, React, Next.js, Node.js, Laravel, PHP, Flutter, React Native, iOS and Android. Corporate sites, ecommerce, marketplaces, and SaaS platforms. Send me your problem I will take care.