I craft autonomous AI workflows — from qualifying inbound leads and voice phone agents to complex API integrations with n8n.
I'm Ansh, a student who got into AI agents and automation almost by accident — I was trying to automate a repetitive task for myself, ended up in an n8n workflow at 2 AM, and haven't stopped since.
Since then I've built several personal projects covering lead qualification agents, voice AI flows with Vapi, and logic-driven automations using the OpenAI and Claude APIs — mostly to understand how these systems actually behave under real inputs, not just in a demo. I care more about a workflow that doesn't break on edge cases than one that just looks good in a screenshot.
I'm still early in this — most of what's below is practice work, not client work — but I'm actively looking to take on my first real builds and learn by shipping something that has to work for someone other than me.
I focus on three core disciplines within the AI and automation space:
I build intelligent conversational agents that interact with incoming site visitors, qualify prospects based on intent and budget, and route ready buyers straight to a calendar or CRM.
I design low-latency voice bots using tools like Vapi, Retell, and ElevenLabs to handle inbound phone screening calls, answer questions, and schedule appointment slots.
I engineer robust self-hosted and cloud n8n workflows that connect CRMs, webhooks, databases, messaging apps, and custom APIs without manual intervention.
System architectures I've designed and built to practice, refine, and test AI agent capabilities.
An automated n8n pipeline that receives incoming webform submissions, uses an LLM to evaluate lead fit against defined criteria, updates HubSpot CRM, and alerts Slack.
A custom Vapi voice agent designed to answer phone calls, ask pre-screening questions, check live calendar availability, and book appointments automatically.
A multi-app automation triggered by Stripe checkout events to automatically provision client Google Drive folders, format welcome notes, and update project trackers.
My 4-step framework for turning complex requirements into resilient automated workflows.
Map out the exact manual bottleneck, data inputs, API endpoints, and desired end outcomes before writing any code or prompts.
Draft strict LLM system prompts, define JSON schema structures, set up error fallbacks, and design the n8n logic flowchart.
Connect webhooks and APIs, test edge-case payloads, verify voice latency, and ensure strict input/output validation.
Deploy to production, configure execution logging, set up error alerts, and refine agent performance over time.
Have a project in mind, want to collaborate, or discuss AI automation? Send me an email directly below.
Feel free to reach out anytime. I aim to respond to all inquiries within 24 hours.