~/kamal-nayanRésumé
AI Engineer || Software Developer

Kamal Nayan
Kumar

I build things with Python, TypeScript, and machine learning — from AI agents that handle real customer calls to a language model trained from scratch. B.Tech in Data Science & AI at IIIT Dharwad, class of 2027.

Kamal Nayan Kumar

About Me

Building AI that actually ships.

I'm a third-year Data Science & AI student who spends summers inside real products. I've helped build an AI receptionist that answers calls and books customers, and a memory layer where AI agents agree on facts before they write.

My work spans the training-to-deployment gap: a 114M-parameter language model trained from scratch, a pipeline that turns a topic into a narrated video, and RAG bots behind real access control. I care about code that reads well and systems that don't surprise you.

run/loss-curve.png

h100 · fp16

e1 · 4.32final · 1.04

steps 1,247

tokens ≈3B

fig 1 · pre-training the 114M-param question-generator

base

IIIT Dharwad — B.Tech, Data Science & AI

admitted

2023 (23BDS026) — graduating May 2027

gpa

8.83 / 10.0

focus

AI engineering · LLMs · full-stack

Experience

Where I've worked.

HobbyFi

May — Jul 2026 · Remote

AI Engineer Intern

AI Receptionist

The agent that answers business calls, texts on WhatsApp, and books customers — one guardrailed copilot across Meta WhatsApp, Plivo, and LiveKit.

Company Memory

A shared memory layer where agents agree on verified facts before writing anything — with permissions per team.

Copilot & Webhooks

Write actions that only fire after vendor confirmation via webhooks, so the agent never persists unverified data.

Learnings: Webhook-driven writes, production guardrails, and shipping under deadlines.

TypeScript · Next.js · Firebase · LiveKit · Qdrant

Buddhi AI

2025 — 2026 · Remote

Backend & Memory Systems

Memory Platform Backend

RBAC, conflict resolution between agents, and a fact registry that governs who can read and write what.

MCP + CLI

Shipped the memory layer as MCP and CLI tools, with OAuth logins into Google Workspace, Notion, GitHub, and Slack.

Learnings: Designing APIs other agents consume, and permission models that survive scale.

TypeScript · Neon · Qdrant · FalkorDB · Resend

Skills

What I work with.

Languages

PythonTypeScriptJavaScriptSQL

AI / ML

PyTorchLlamaIndexRAGLLM fine-tuningComputer Vision

Frameworks & Web

FastAPINext.jsReactTailwind CSS

Data & Storage

PostgreSQLQdrantFalkorDBFirebase

DevOps & Tools

GitDockerGitHub ActionsVercelRenderLinux

AI Tools

ClaudeChatGPTGeminiCursorNotebookLM

Contact Me

Let's talk.

Open to 2027 internships and grad roles in AI engineering — plus freelance builds and interesting side projects.

opens your email app — or write to kamalnayanofficialwork@gmail.com directly