Founder story · www.mlops.dev

Building the infrastructure
edge AI deserves.

Raghunathareddy GR built MLOps.dev to solve a real problem — deploying ML models to edge devices is broken. SSH scripts. USB sticks. Zero visibility. One founder. One obsession. One platform to fix it all.

Founder and CEO of MLOps.dev
CEO & Founder · MLOps.dev
Bengaluru, India · www.mlops.dev
The founder story

Why I built MLOps.dev.

It started with a factory in Bengaluru. A computer vision model — months of training, nights of tuning, 94% accuracy on the benchmark — deployed to 12 industrial cameras via a Python script someone wrote in an afternoon. Three weeks later, no one knew what version was running on which camera. Drift had crept in on six of them. Nobody had an alert. Nobody even knew to look.

"I spent more time managing deployments than improving models. That's when I realised the infrastructure was the real problem — not the ML."

I'm a builder with a background in ML engineering and systems design. I've shipped models to edge hardware across manufacturing, retail, and healthcare. Every single time, the operational layer was an afterthought — bash scripts, cron jobs, and prayers that the device would reconnect to sync the update.

The cloud-native MLOps tools — great for data centres, useless for the edge. Kubeflow doesn't run on a Jetson Nano. MLflow doesn't care if your device goes offline for six hours in a factory basement. The edge has fundamentally different requirements, and nobody was building for them.

So I started building MLOps.dev — not as a side project, but as the company I wish had existed when I needed it. An open-source agent small enough to run on any ARM board. An offline-first sync protocol that works at 2G speeds. A drift detection system that fires before your users notice. Infrastructure that meets the real world where it actually is.

Founder timeline
EARLY YEARS
First lines of code, first obsession
Started programming as a teenager. Fell in love with systems — how things actually work at the hardware level. Built embedded projects before knowing what ML was.
ML JOURNEY
Training models, hitting production walls
Trained and deployed computer vision and NLP models across real industry projects. Discovered the hard way that model accuracy is 20% of the problem — operations is the other 80%.
THE TURNING POINT
The Bengaluru factory moment
Deployed a model to 12 industrial cameras. Three weeks later — unknown versions, silent drift, zero monitoring. Spent a week recovering what should have taken minutes. Decided to fix this properly.
VALIDATION
20+ customer discovery calls
Talked to ML engineers at manufacturing, retail, and healthcare companies. Every single one had the same problem. The market gap was clear and no one was filling it.
NOW · 2025
Raghunath building MLOps.dev full-time
Open-source edge agent live. Backend API shipped. 50+ teams on the waitlist. First design partners onboarding. Building from Bengaluru, for engineers everywhere.
"The best ML model is the one that's actually running in production — not the one that scored 0.001 better on a benchmark nobody ships."
Raghunathareddy GR · CEO & Founder · MLOps.dev · www.mlops.dev
Mission & values

What MLOps.dev stands for.

Three principles that drive every decision — from the agent binary size to the pricing model.

Real world first
Every feature is designed for the edge — where internet is spotty, hardware is constrained, and failures have physical consequences. We build for 2AM in a factory basement, not the ideal demo environment.
Open by default
The edge agent is open-source and always will be. ML teams deserve to audit every line that runs on their hardware. We monetise the cloud control plane — not the agent. Transparency is the foundation.
Drift is the enemy
Silent model degradation is the biggest unaddressed risk in production ML. We treat drift detection as a first-class citizen — not an add-on. Every deployment includes monitoring, not as an optional extra.
Core values
01
Honesty over hype
We don't claim to manage 10,000 devices on our homepage until we've actually done it. Every stat we publish is real. The 50+ on our waitlist counter is a real API call.
02
Engineers first
Our users are ML engineers, not executives. The product is designed for people who read stack traces for fun. No enterprise jargon. No "synergy". Just tools that work.
03
Offline is normal
We treat offline as the default state, not an edge case. Every feature works without internet. Connectivity is a bonus. The device should always keep running.
04
Ship then learn
We'd rather have 10 design partners using an imperfect product than 100 people on a waitlist for a perfect one that doesn't exist. Build in public. Fix fast.
50+
Teams on the waitlist
↑ growing daily
<8MB
Agent binary size
any ARM device
20+
Customer discovery calls
before writing line 1
1
Founder, full-time
Bengaluru → world

Want to build with us?

Looking for design partners, early users, and engineers who care about edge ML. Let's talk.