Hi, I'm Naman Agarwal

Senior Software Engineer specializing in AI Agents, Machine Learning, and High-Performance Systems. Currently building autonomous LLM agents and agentic payment infrastructure at Juspay.

naman@workspace:~

git:main

NORMAL about_me.md

utf-8 1,1 All

Console

visitor@naman-os:~$

whoami

visitor@naman-os:~$ whoami

Naman Agarwal

I am a Senior Software Engineer at Juspay specializing in AI Agents, LLM Orchestration, and High-Performance Systems. I design and scale agentic workflows, custom Model Context Protocol (MCP) toolkits, and robust backend payment infrastructure.

My engineering approach is bottom-up and first-principles-driven: I build autonomous agent pipelines, custom LLM evaluation frameworks, and high-throughput Rust services from scratch. I focus on bridging the gap between advanced AI reasoning models and production-grade systems engineering.

I served as the Tech Lead for the Android Club at VIT Chennai during my B.Tech, where I organized hackathons and mentored juniors in clean development practices. I thrive at the intersection of AI research, systems programming, and product execution.

neofetch --skills

visitor@naman-os:~$ neofetch --detailed-skills

AI, Agents & Machine Learning

LLMs (Claude, GPT-4)

AI Agents & Workflows

Model Context Protocol (MCP)

RAG & Embeddings

Federated Learning

FaceNet CNN / OCR

Prompt Optimization

Languages

Rust

Haskell

Go

Python

TypeScript

JavaScript

Java

C++

Back-End & Frameworks

Node.js

Express

React

React Native

Flask

GraphQL

Redux

DevOps, Cloud & DB

Docker

AWS Lambda / ECS

PostgreSQL

MongoDB

Hasura ORM

Firebase

Git

git log --experience

visitor@naman-os:~$ git log --graph --oneline --experience-tree

Juspay SDE II (AI, Backend, & Frontend)

2022 - Present

Godel AI (Scrapers): Designed an LLM selector generator prompting models for DOM selectors, exposed to a dashboard for approval. Implemented a health-scoring system to monitor scraper performance and stagger traffic dynamically for self-healing.
Agentic Payments: Built MCP (Model Context Protocol) servers to expose banking APIs (auth, payment methods, product discovery) to agents. Created a RAG integration bot using documentation to debug SDK integrations.
Backend & Crypto: Contributed to backend payment systems (Euler), decoupling AuthN/AuthZ journeys. Deployed cryptographic modules (ECC, JWE, shared secrets) on a high-throughput Rust-based 3DS authentication server.
Custom UI Framework: Developed core widgets (Dropdown, Tooltip, LazyScrollView) in PureScript for Juspay's custom cross-platform engine, optimizing rendering via virtual DOM diffing, Java reflection, and JS OTA updates.

Rust PureScript Haskell Python LLMs MCP Functional Programming Cryptography

ZeeQ Backend Engineer

2021 - 2022

Architected highly scalable back-end microservices for data archiving using Clean Architecture principles.
Authored over 40% of the backend codebase, including secure authorization and RBAC middlewares.
Engineered PostgreSQL relational models and configured schema triggers with Hasura ORM and GraphQL endpoints.

Node.js Express GraphQL Hasura PostgreSQL Docker

docker ps --projects

Baasha (AI Video API & MCP Server)

A Python-based Flask API and custom MCP server orchestrating collaborative AI agents (Creative Director, Production Architect) built using Google's Agent Development Kit (ADK) and Gemini Flash. Integrates Human-in-the-Loop review gates, dynamic Living Bibles (pgvector RAG) for visual/character consistency, ElevenLabs for voice synthesis, Replicate (Stable Diffusion/Kling) for video frames, and MoviePy/FFmpeg for automated media pipeline compilation.

Python Google ADK MCP (FastMCP) pgvector Replicate API ElevenLabs

Demo

Family Tree Builder

Graph parsing engine that maps relational multi-generational graphs from electoral tables. Integrates custom CAPTCHA solvers, Tesseract OCR for document scans, and Scrapy to parse 25,000+ data nodes.

Python Tesseract OCR Scrapy NetworkX

Source

Invoice Central

Web app designed for offline store billing, listing products and compiling invoices instantly. Optimized queries using complex MongoDB Aggregation pipelines, reducing page load latency by 5%.

Node.js Express MongoDB React

Source

Confab Video Chat

Full-duplex real-time video chat solution using custom WebSockets signaling and WebRTC protocol. Leverages dynamic mesh connections to support robust performance and tree-based data buffers.

JavaScript WebRTC WebSockets Node.js

Source

Face Recognition IoT Gateway

Automated attendance verification system leveraging Google's FaceNet CNN model for face embedding extraction. Configured MQTT brokers for real-time synchronization between edge IoT gateway devices and a centralized PostgreSQL database.

Python FaceNet MQTT IoT Gateway

Source

RN Transition View

Highly performant open-source library built on top of Reanimated 2 to compute matrix layouts on UI threads, delivering complex transition animations at a fluid 60fps.

React Native Reanimated 2 TypeScript

Source

cat achievements.txt

Survival Startup Title

Awarded 'Survival Startup' at the UBC Hackathon for high resilience engineering.

VIT Android Tech Lead

Served as Tech Lead for VIT Chennai Android Club, organizing local workshops and events.

State R.O.L Recommendation

Awarded official Letter of Recommendation from Additional Director General of UP Police, India.

Play Store Deployment

Architected and published full-stack productivity applications, managing client updates.

Federated Learning Research

Completed a Research Internship in Federated Machine Learning at VIT Chennai.

ssh contact@naman.dev

visitor@naman-os:~$ ssh contact@naman.dev

Email

namanddev7@gmail.com

Phone

+91 9651111184

Location

Bengaluru, Karnataka, India

Networks