Harsh K.

I'm an AI Automation Engineer specializing in RAG, AI agents, and knowledge systems. I build production RAG applications, AI-powered product and website documentation, document Q&A systems, and internal knowledge assistants, and connect them to business workflows with n8n and APIs. I specialize in Agentic RAG and AI automation, building knowledge bases, AI search systems, document assistants, AI agents, and automated workflows using n8n, Pinecone, Supabase, OpenAI, Claude, and langchain and Vercel AI SDK. Whether you have a SaaS documentation site that needs an AI assistant, a large collection of business documents, an internal knowledge base, or recorded content that needs to become searchable, I build the RAG pipeline that turns that information into something your users or team can actually interact with. 🥇 n8n Certified — AI, Testing & Best Practices 🥇 n8n Certified — Integrations: APIs & Connected Workflows ⭐ 100% Job Success Score ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ ⚫ RAG & KNOWLEDGE SYSTEMS ➡️ Custom RAG pipelines using Pinecone, Supabase/pgvector, Qdrant and other vector databases ➡️ Document Q&A systems for PDFs, DOCX, TXT, CSV and other business documents ➡️ Internal knowledge assistants connected to Google Drive, Dropbox, Slack and other business tools ➡️ Product documentation & AI search for websites, help centers and knowledge bases ➡️ Multi-source RAG combining documents, websites, databases, APIs and structured business data ➡️ Video & audio RAG, transcription, timestamped knowledge extraction, chunking and semantic retrieval ➡️ RAG agents that combine retrieved knowledge with tools, APIs and business workflows ➡️ Source citations, metadata filtering, semantic search, reranking, conversation memory and document re-indexing 🔵 n8n & AI AUTOMATION ➡️ Automated document ingestion and knowledge-base synchronization ➡️ API integrations, webhooks and event-driven workflows ➡️ RAG pipelines that automatically process new, updated and deleted content ➡️ AI agents connected to CRMs, databases, internal tools and external APIs ➡️ Lead qualification, routing, appointment booking and follow-up automation ➡️ Complex workflows with branching, conditional logic, error handling and data transformation 🟣 AI AGENTS I build agents that don't just generate text — they retrieve the right information and take action. ➡️ Knowledge-grounded AI agents ➡️ Tool calling and API integrations ➡️ RAG + n8n agent workflows ➡️ CRM-connected agents ➡️ Customer support and internal operations agents ➡️ OpenAI / Claude-powered conversational systems 🟢 THE TYPES OF SYSTEMS I BUILD • Internal company knowledge assistants • Slack knowledge bots • Chat-with-your-documents applications • Product documentation AI search • Customer support RAG agents • Multi-source enterprise knowledge bases • RAG + n8n automation systems • Agentic RAG applications • AI search over business data • Video/audio knowledge assistants ⚙️ MY APPROACH I don't treat RAG as simply “put documents into a vector database.” A production system needs the complete pipeline: Ingestion → extraction → chunking → metadata → embeddings → vector search → retrieval/reranking → LLM → citations → actions I focus on clean architecture, reliable ingestion, accurate retrieval, proper source attribution, error handling and systems that can actually be maintained after handoff. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ ⭐ RECENT RAG & AI AUTOMATION WORK ▪ Built a RAG-powered AI agent using n8n + Supabase Vector Store for automated document retrieval and Q&A ▪ Built n8n workflows connected to vector stores for knowledge retrieval and AI automation ▪ Developed an AI SDR system automating lead capture, qualification and multi-channel engagement ▪ Built AI voice appointment automation using Vapi + n8n + Make ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ WHY CLIENTS HIRE ME I work as a long-term technical automation partner, not someone who simply delivers a workflow and disappears. You own the workflows, credentials and infrastructure. I build systems that are documented, understandable and designed to be extended when your requirements grow. If you already have documents, data, a knowledge base, an existing automation stack, or an AI workflow that isn't working reliably, send me what you're trying to build. I'll tell you what the architecture should look like and what actually needs to be built.