life @ google
I’m glad to announce that it has been six months since I started my career at Google, as a research engineer at Google DeepMind.
A detailed blog about my preparation is on the way, but today I want to talk about what the title suggests: life at Google!
So it all began when I received my offer and I reached the office for the first time on November 10, 2025. Before I knew it, I was introducing myself to colleagues and seniors. I got a taste of the fast-paced culture at Google immediately as I was assigned to my first project: the information retrieval suite for Gemini 3.1 Deep Think!
I still remember the conversation I had with my manager Paul regarding the project. The first task was a little underwhelming: I had been assigned a list of websites that I had to crawl and retrieve data from. At a glance it looked quite menial but he assured that I would soon be asking for help. I was looking forward to the challenge.
It took me five days to realize that it was, in fact, quite menial.
After navigating hundreds of pages and manually filling up a Google Form for every piece of data I found relevant, I was tired.
But just two weeks in, I was done earlier than designated! My manager was happy, ready to assign a real task to me.
I read the statement for the next task: I was assigned a list of websites that I had to crawl… “Huh,” I thought to myself, “isn’t this the same thing?” I started working anyway. Maybe I was missing something.
I approached one of my coworkers and asked her about this.
“Hey,” I said. “Can I ask you something?”
She looked up from her screen. “Sure.”
“Do you know what exactly we’re building? Like… End to end?”
“Um. Building?”
“Yeah, the research. I’ve just been collecting data for five months.”
“Oh. That. Well, don’t worry about it. You’ll get your credit.”
“That’s not what this is about…” I scratched my head. “When do I get to help with actual research?”
“What do you mean?”
“Like… The part where we design things. Or analyze results.”
She thought for a moment.
“I think that’s mostly senior engineers or research scientists,” she said. “We don’t really get that kind of scope yet.”
I went back to my desk. Maybe I could talk to a research scientist.
I asked Paul to connect me to our lead, Dr. Luca.
“Any particular reason?” He asked. “Dr. Ed is overutilized at the moment.”
“I just want to understand the system end to end,” I said.
“Bring it up in the next appraisal lunch,” he said.
I brought it up in the next appraisal lunch.
We were seated in a room that was not marked differently from any other meeting room. The calendar invite simply said: Appraisal Lunch Q1 Review.
I had the impression that Dr. Luca would be present but I was wrong.
I waited for an appropriate moment and spoke.
“Actually, I wanted to talk to Dr. Luca. I have some questions regarding research.”
“Oh yeah.” One of my colleagues said. “Paul mentioned that you do. Since you’re off your probationary period we can finally give you access to him.” He paused for a moment before continuing. “Since lunch is almost over, let’s also review the paper.”
A few chairs shifted slightly as people turned toward the screen in the room. It displayed a new Deep Think paper discussing the latest benchmarks. My heart skipped a beat seeing my name right in the middle of the author list.
I was excited for a moment but then I remembered that I hadn’t worked on any paper all this time. Moreover, I couldn’t find Dr. Luca’s name anywhere. I looked around frantically but then a hand stopped me. It was Paul’s: his expression clearly said, now’s not the time.
I was quite blank after that, so I approached him later.
“What was that about?” He demanded.
“What was what about?”
Paul let out a long, long sigh. “Just. Let’s get you to talk to Ed.” He opened his laptop.
On his screen were the letters ‘back at it, Paul’.
“Um… Paul?” I said. “This is…?”
Suddenly the laptop blurted out:
It looks like your message might be incomplete – what’s on your mind?
Dr. Ed sounded like… a woman. Huh.
“Can I ask you something?” I said. Paul looked up from the monitor. “This is just Claude, isn’t it?”
He read my confused look and closed the lid. “I know exactly what’s going on. I should’ve figured it out weeks ago.” He took a pause and poured himself some water. “So you didn’t read the code of conduct.” I felt my face flush a little. “Who even reads things these days, right?”
I was a little mad now because I just wanted to talk to Dr. Ed.
“But what about Dr. Ed?” I asked, calmly.
“What about… You just talked to him?” Paul said. I had never felt stupider.
I got back to my desk after that and read the code of conduct. Whoever wrote it should be in jail, but I digress. My head was spinning – suddenly everything made sense.
I started noticing it everywhere after that.
Dr. Ed, after all, isn’t a man. He isn’t even a person. As per the code of conduct: ‘All research leads at Google DeepMind are assisted by or instantiated as large language model clusters. Interaction with research leads is subject to the AI usage policy.’
“Ed Luca” is, by now you must’ve guessed, an anagram. I wasn’t so quick to notice.
I spent the week discovering more.
Apparently, nearly all research done in big tech is led by ML ‘experts’ but they’re all just LLMs. Internal model clusters generate experiments, benchmark them, reason about current approaches, and try to optimize them – everything done a hundred times faster. Every single researcher that we know is simply a data entry guy. Honestly, that’s kind of nice. I hate thinking.
So yeah. Ed? A clanker. Most researchers you’ve heard of? They’re just using clankers. Andrew Ng? Is a model. I know. I wonder who else isn’t real…
Needless to say I was called in for a review soon after my conversation with Paul.
As of today, I have been laid off from Google for underutilizing their research infrastructure. I did sign a non-disclosure agreement about the LLM usage, but who’s going to read this blog anyway? Waiting for further opportunities though! #cfbr