I quit Meta and left a comfortable salary to build a startup. Leaving wasn't easy, but now my AI receptionist answers millions of calls.

• Ruchir Baronia left Meta to launch Frontdesk, an AI startup that automates business calls.
• His early coding projects, including Rescuer, showed him the real-world impact and reach of code.
• Now, he works with a team of other former Big Tech professionals to help businesses prevent customer loss.

My clearest memory from middle school is pacing in my bedroom, phone in hand, repeating the word 'wake' again and again. I was testing a voice app I had written. If I said the phrase just right, the app would respond. If not, I would tweak the code and try again.

I had just learned to code from YouTube. My apps were getting downloads, and I was addicted. I would run home from school, drop my backpack, and open the reviews before starting my homework. It was the first time I saw that code written alone in my bedroom could reach people I would never meet.

After studying engineering and business at UC, Berkeley, and spending nearly two years working at Meta, I'm now running Frontdesk, an AI startup that helps businesses automate millions of conversations with AI.

When I was starting out, I realized code could protect people

My early programming projects were playful. I built a text-to-speech app and a speech-to-text app. Then, one of my relatives slipped and fell at home. He was yelling for help, but nobody was home, and his phone was out of reach. If his phone had been able to hear him, it could've brought him help immediately.

In 2016, I developed Rescuer, a voice app that enabled users to trigger an alarm by shouting a secret phrase, automatically sending their location, photos, and audio to designated emergency contacts.

Rescuer earned recognition from political officials and the press. What mattered more was that I learned that code written from my bedroom could help someone. From then on, I began building for real impact.

I started answering questions on coding forums, recording YouTube coding tutorials, and writing for an engineering blog. Over time, my answers on forums like Stack Overflow reached millions of developers, which only deepened my passion for continuous learning.

Living in two worlds

There was still one thing I couldn't figure out, no matter how many downloads my apps received. I didn't know how to turn an app into a real business. I didn't understand pricing, customers, or why some products became companies while others remained weekend projects.

As I was deciding where to attend college, I knew I wanted to learn about the business side.

I chose UC Berkeley's M.E.T. program because I could study both engineering and business. I graduated with an EECS (electrical engineering & computer science) and business degree in three years.

The experiment

In 2023, when ChatGPT had its big moment, I asked myself, What if I could just talk to this thing on a real phone line?

Over a weekend, I hacked together one of the first prototypes that connected ChatGPT to a real phone number. I posted a short clip and a tweet, and it got a lot of attention.

The tech was not ready to be a real product. Latency was bad. Reliability was fragile. The models made strange errors. But that experiment planted a question in my mind that I could not shake: If this ever became good enough, what would it look like to use it to help real businesses?

Leaving comfort

By graduation, I had offers to work in finance at Citadel or in engineering at Meta. I'd spent all of college living in both worlds, and I didn't want to pick just one side, but I chose Meta.

I shelved my phone experiment to join Meta. It was a dream setup. My team was incredibly supportive, I was learning how to build at a massive scale, and life was good — from the smart colleagues to the famous free food.

I was part of a fintech team where I'd push one line of code, and millions of transactions would be affected. It felt less like a corporation and more like a high-growth startup; I was given incredible ownership from day one and learned more in almost two years than I expected to in five. The scale was exhilarating.

But I was spending all my time behind the code. The business side, the part I'd learned to love at Berkeley, felt far away. As an engineer at a big company, there was no clear path back to the center.

I kept hearing about AI in the hallways

I wasn't working directly on AI, but I could sense the momentum building. Every few months, the models improved noticeably. My ChatGPT phone hack stopped feeling like a viral stunt and started to feel like a glimpse of the future.

After my post went viral in college, hundreds of businesses reached out, asking for a solution. I kept thinking about all those businesses, where a missed call meant a missed customer. I knew there was demand, I knew nobody was serving them, and I knew I wanted to be the one to build it.

Leaving wasn't easy. Meta was comfortable, offering an incredible salary, stock refreshers, and interesting problems. Everyone told me to stay another year, vest more equity, and build more credentials. The rational move was to wait.

I kept asking myself, if I waited, would I regret it? The window felt finite. Every month I stayed was a month these businesses were not being served.

I quit my job at Meta in February 2025 after 1 1/2 years at the company

I raised capital, moved to NYC from California, and began building Frontdesk. I'm now the founder and CEO.

Frontdesk is a comprehensive AI operating system that engages with customers across all channels and performs real-world tasks, such as scheduling and follow-ups.

We work in-person in an office in SoHo. Most people in our office share a similar story to mine; they left Microsoft, Amazon, or Meta, walking away from good salaries and stock options, because they believed this was worth building.

I feel lucky to work with people who push me to be better every day.

Still pacing in my room

Most nights, my work looks the same as it did in middle school. I pace, talk into a phone, and listen for what feels off. The difference is that now, when it works, it works across millions of calls. It's the same instinct that led me to build Rescuer years ago.

The phones are still ringing, and the models are still getting better. I know where I want to be while that happens: close to the code and building something worth leaving comfort behind.

Did you quit a Big Tech job to build your own company? Email this editor at lhaas@businessinsider.com.