AI Automation Engineer  ·  Agentic AI with Claude  ·  MCP  ·  n8n Verified Creator

I'm an AI Automation Engineer specializing in building production-grade agentic AI systems, LLM pipelines, and automation platforms that solve real business problems — not just demos. My work spans the full stack: from document ingestion and AI processing to workflow orchestration and business output delivery.

• 🧠 Currently: building a "Company Brain" — agentic AI systems with Anthropic Claude, Claude Code subagents & MCP that give CFOs and leadership teams unified AI visibility into financials, CRM, operations, and workflows
• Specialized in AI agents (Claude, MCP, LangChain, CrewAI), RAG systems, document intelligence, and end-to-end workflow orchestration
• Experienced in fintech automation integrating APIs such as Equifax, compliance services, and CRMs
• Focused on shipping systems that go from prototype to production, reliably and at scale

Build real systems. Automate everything. Scale what works.

Self-hosted multilingual phone voice agent - LiveKit Agents + Sarvam STT/TTS + OpenAI + SIP

• ~700ms-1.2s round-trip latency, under Rs 5.5/min
• English + 5 Indic languages; demo: real-estate site-visit booking agent

voice-agent-starter-kit

An Obsidian vault + any AI agent (Claude Code, Cursor) = a curated second brain for your business

• /onboard interviews you, researches your tools' APIs & MCP servers, generates your instance
• Plain markdown, MIT license, zero lock-in

company-ai-os

Deep-research agent platform with auth, billing, and multi-step agent orchestration

• Next.js + Supabase + Stripe + Tavily - live product

Live app | GitHub

Select text in any app, press Ctrl+Shift+Space - it becomes a well-structured LLM prompt

Prompt-Kitcha

End-to-end document processing and automation using Azure AI + Power Automate

• Processed 20+ financial document types using OCR + LLM pipelines
• Integrated Equifax APIs for credit and identity data retrieval
• Built Zoho Creator workflows for deal automation and CRM sync
• Reduced manual processing effort by ~30%

Architecture: Input → Power Automate → Azure OCR → LLM → Validation → APIs → CRM

🔗 Azure Document Intelligence  |  Power Automate Flows

Multi-step automation pipelines for compliance and credit data

• Designed full pipelines: OCR → LLM → Validation → CRM → API calls
• Converted XML API responses to structured JSON for downstream use
• Integrated Equifax, ACN, PPSR, and title check APIs
• Implemented retry logic, conditional flows, and fault-tolerant error handling

AI agent workflows published as a Verified n8n Creator

• Built agents for lead generation, job scraping, and social media automation
• Integrated OpenAI + Apify + external APIs into fault-tolerant pipelines
• Reduced manual effort by ~50% across use cases

🔗 n8n Creator Profile

Conversational search system built on Retrieval-Augmented Generation

• Built with LangChain + Mistral + ChromaDB
• Improved document search accuracy by ~35%

🔗 GitHub Repository

Natural language to SQL interface for non-technical business users

• Enabled plain-English database queries without SQL knowledge
• Powered by LLM-based query translation and validation

🔗 GitHub Repository

Template-driven document automation with backend integration

• Automated generation of structured documents from dynamic data inputs
• Integrated with backend systems for real-time output

🔗 Project Details

System-wide prompt engineering tool — transform any text into a structured LLM-ready prompt with a single keystroke

• Select text in any application, press Ctrl+Shift+Space, and instantly receive a well-structured prompt
• Works across all apps system-wide — browser, IDE, notes, documents
• Designed for developers and power users who work with LLMs daily
• Eliminates prompt iteration overhead at the point of input

🔗 GitHub Repository

Full-stack AI meeting intelligence platform — transcription, summarization, and sprint management

• AI-powered meeting transcription with automatic action item extraction
• Sprint backlog generation and prioritization from meeting context
• Built with Next.js + OpenAI + real-time database for live collaboration
• Live deployment serving active users

🔗 GitHub Repository  |  meetingsai.netlify.app

Modular skill system for Claude — reusable, composable instruction sets that extend Claude's capabilities across workflows

• Architected a skill-based framework enabling Claude to consistently execute complex, multi-step tasks
• Skills cover domains including document creation, data analysis, frontend design, and automation workflows
• Designed for composability — skills stack and combine across pipelines
• Enables rapid deployment of Claude into production workflows without prompt re-engineering

🔗 GitHub Repository

Multi-agent system for autonomous evaluation and scoring of academic research papers

• Orchestrates specialized agents across methodology review, citation analysis, and result validation
• Produces structured evaluation reports with scoring rubrics and improvement recommendations
• Built for researchers, reviewers, and institutions processing high volumes of papers
• Demonstrates agentic coordination patterns: planning → delegation → synthesis

🔗 GitHub Repository

• AI-powered document processing systems — OCR + LLM + validation pipelines
• RAG-based search and chatbot systems — semantic retrieval over enterprise data
• AI agents for automation — lead generation, scraping, multi-step orchestration
• API-driven automation pipelines — CRM, fintech, compliance workflows
• Real-time AI applications — full-stack, backend + frontend integrated

• Portfolio: dinakar.vercel.app
• LinkedIn: linkedin.com/in/dinakar-selva
• Upwork: upwork.com/freelancers/dinakar
• Email: dinakars2003@gmail.com
• n8n: n8n.io/creators/jamesdinakar
• Medium: medium.com/@dinakars2003