Nilesh Pethani

Nilesh
Pethani

I lead AI strategy and transformation at iMocha. I build AI agents, cloud platforms and data systems that hold up in production.

Inspired by the ESP32 sound visualizer I built at home.

About me

I have spent more than thirteen years at iMocha, a skills assessment and talent analytics platform. I started deep in databases and web applications, and today I lead work across engineering, AI/ML and cloud infrastructure.

My job is to turn AI from an idea into something people use every day: agents that answer real questions, platforms that scale, and data that can be trusted.

Role
Associate Director, AI Strategy and Transformation
Company
iMocha
Experience at iMocha
13+ years
Based in
Pune, India

What I work on

AI agents and LLM systems

Agent pipelines that understand skills, jobs and learning content. Natural-language questions turned into safe, explainable answers. Knowledge graphs that map how skills relate to each other.

LangGraph, Python, FastAPI, PostgreSQL + pgvector, Neo4j

Cloud infrastructure

Platforms on Azure and AWS, from app hosting and containers to API gateways and multi-account setups. Built to be secure, observable and simple to run.

Azure App Service, Container Apps, AKS, API Management, AWS ECS Fargate, RDS, Terraform

Data and databases

Where I started, and still close to my heart. High-availability SQL Server, large schema and data migrations, and search at scale.

SQL Server (Always On), PostgreSQL, Elasticsearch, ClickHouse

Application modernization

Moving long-running products to modern stacks step by step, without stopping the business.

ASP.NET, .NET, Angular

How I like to build

After hours

Jarvis

My personal AI assistant, living in Telegram. It is built with LangGraph and Azure AI Foundry and uses a three-tier memory model, so it remembers what matters and forgets what doesn't.

It started years ago as a small voice assistant on an ESP32 board.

  • ESP32 sound visualizerA frequency-band equalizer that reacts to music.
  • NTP clock and weather stationSmall ESP32 devices that stay in sync with the internet.
  • MagicMirror² displayA smart mirror running on a Raspberry Pi 4.
  • Rechargeable reading lightA 12V LED lamp with an 18650 battery and TP4056 charging circuit.

Let's talk

Happy to talk about AI in HR tech, cloud architecture, or a fun hardware project.

hello@nileshpethani.com