I build production AI systems — agent runtimes, orchestration layers, and the infrastructure that keeps them reliable. My work sits at the intersection of TypeScript, LLM tooling, and systems that have to run in the real world: memory, tools, cost controls, and human oversight.

Currently shipping Agentium, a TypeScript-native framework for multi-agent systems on Node.js.

const engineer = {
  name: "Akash Sengar",
  role: "Engineering Leader",
  company: "Xhipment",
  focus: ["AI agent systems", "TypeScript", "open source infrastructure"],
  building: "Agentium — TypeScript-native agent orchestration for Node.js",
  contributing: "agno-agi/agno",
  education: "Computer Science, VIT Vellore",
} as const;

A TypeScript-native agent orchestration framework for Node.js. Model-agnostic, batteries-included, and designed for production — not demos.

import { Agent, openai } from "@agentium/core";

const researcher = new Agent({
  name: "researcher",
  model: openai("gpt-4o"),
  instructions: "Research thoroughly and cite sources.",
});

const result = await researcher.run("Summarize Q4 market trends");

Production concerns are first-class: cost tracking, semantic cache, audit trails, sandbox execution, and multi-tenant isolation.

Documentation · GitHub · npm · Website

agno-agi/agno  

Contributor to one of the most widely used open-source agent frameworks — features, fixes, and production-oriented improvements.

I also publish occasional walkthroughs on agent architecture, TypeScript internals, and system design on YouTube.