Kulbhushan V.

AI Agents & RAG Developer | Enterprise AI Automation

I build AI agents, voice agents, and RAG copilots that run in production for companies like Johnson & Johnson, Coca-Cola, and Panasonic. Top Rated Plus, 100% Job Success, 18+ years. Most AI projects die between the demo and the deploy. Mine don't, because I build the unglamorous parts that make AI trustworthy: permissions, human-approval gates, audit logs, and evaluation, then document everything and hand it over so your team runs it without me. AI AGENTS I BUILD - Sales & Marketing Agents — inbound lead qualification on WhatsApp, web, and email, 24/7, synced to your CRM (HubSpot, Salesforce). Outreach drafting, campaign analysis, and content workflows in your brand voice. A recent build cut a hospitality group's manual lead qualification time by ~50%. - Voice AI Agents — always-on conversational agents for sales, IT helpdesk, and HR queries. Phone and chat, natural language, escalation to humans at the right moment. - HR & Talent Agents — AI interview and candidate-assessment agents with structured scorecards, plus onboarding and offboarding automation (accounts, checklists, first-week plans, knowledge-transfer exits). Explainable, bias-aware, built for HR teams, not just engineers. - CFO & Finance Agents — automated reporting and reconciliations, expense and margin analysis, SKU/product profitability, on-demand close support. Every figure traceable to source, every action logged, built for auditors to love. - AI Knowledge Bank (RAG) — "chat with your company docs": permissioned copilots over your contracts, SOPs, and drives, with source-cited answers and role-based access. No hallucinated policy answers. - Custom Workflow Agents — agentic AI with tool use, function calling, and multi-agent orchestration on Claude and OpenAI, wired to your real systems (WhatsApp Business API, HubSpot, QuickBooks, Slack, Twilio, your APIs) via n8n, Make, or Python. Beyond ChatGPT, custom agents on your data, with guardrails. If it's repetitive and rule-describable, it's automatable. Underneath all of it: the data engineering and BI (Azure, AWS, Snowflake, Databricks, Power BI, Tableau) that make AI outputs accurate. RECENT WORK • Claude-powered WhatsApp sales agent, multi-location hospitality group — 24/7 lead qualification, human approval on every reply, HubSpot sync. • AI evaluation engine scoring procurement bids on compliance, technical, and commercial criteria — every score traceable to its source clause. • Executive AI dashboard unifying ERP/CRM data — forecasts, anomaly alerts, and plain-English narratives for the C-suite. • Industrial decision-support platform — multi-system readings into one asset-health record, deterministic severity engine, AI-written summaries, staff approval gate. WHY U.S. CLIENTS WORK WITH ME Daily overlap with U.S. business hours, proactive written communication, and delivery for American enterprises, including Johnson & Johnson, Coca-Cola, Primo Water, Paxful, Target, and Ceribell (healthcare). Comfortable with security reviews, data boundaries, HIPAA/SOC2-sensitive environments, and procurement processes. What clients say ⭐⭐⭐⭐⭐ “Five stars all the way… great attitude and great attention to detail. Very professional.” ⭐⭐⭐⭐⭐ “They took an enormous task and completed it amazingly… a pleasure to work with.” ⭐⭐⭐⭐⭐ “Understood our business almost better than ourselves… communication was top-notch.” 🛠Tools🛠 Claude (Anthropic), Claude Code, OpenAI GPT, RAG, voice AI, LangGraph, n8n, StackAI, Make, Zapier, Python, PySpark, SQL, AWS Bedrock, Azure AI, GCP Vertex AI, Snowflake, Databricks, Power BI, Tableau, HubSpot API, WhatsApp Business API, Twilio