About

I build AI products from operating idea to deployed system.

I'm Harshitha Chittapragada, an AI Product Engineer and Founder of Agent Mag. I work across product discovery, roadmap definition, agent architecture, implementation, integrations, deployment, and the operational details that determine whether a system remains useful after launch.

At Agent Mag, I lead an 86-page product, 62 API routes, a 492-skill open ecosystem, Agent Mag AI, and a governed automation runtime for finance and broader operations. I also designed and operate Jarvis, a master-agent system with Email, Calendar, and Meeting specialists. My enterprise hiring work covers Microsoft Graph intake, candidate matching, resume attachments, duplicate-reply prevention, restart recovery, connection pooling, and Azure deployment. Content automation and OpenClaw and Hermes client deployments extend the same work into publishing and private agent operations.

Claude Code and Codex are core implementation tools in that process, but the job is larger than prompting. I define the operating model, choose the system boundaries, connect providers, inspect failures, ship the product, and keep the human decision points clear where an automated action has consequences.

How I work

AI proposes. Systems control.

Models are useful for extraction, classification, drafting, and investigation. Deterministic services should own permissions, execution, retries, and consequential actions.

Recovery is a product feature.

A workflow is not reliable because it succeeded once. It needs idempotency, persisted state, retry behavior, observability, and a clear path through partial failure.

Human review belongs in the architecture.

Publishing, sending, approving, and changing business records need explicit boundaries. Review is not friction when it protects trust.

Background

Engineering foundation, operator's perspective.

My formal training is in Artificial Intelligence and Data Science. Building products across agent infrastructure, recruitment operations, publishing, and client implementation taught me to connect code with the people and processes around it.

I work across TypeScript, Next.js, Python, Java, PostgreSQL, Microsoft Graph, Azure, OpenClaw, Hermes, n8n, workflow runtimes, and model APIs. I'm comfortable moving between product discovery, implementation with Claude Code and Codex, debugging, deployment, and operational follow-through.

I'm deliberately growing toward enterprise-grade architecture: multi-system products with explicit security boundaries, durable state, observable execution, failure recovery, and human control. I want to deepen the technical judgment required to own increasingly complex systems end to end.

I'm looking for roles in enterprise product engineering, automation platforms, AI operations, or technical implementation where I can deepen that responsibility against real business constraints.

Explore the systems I've worked on