Erran Berger

Mostly theirs

Claims

• Owning the stack end to end lets LinkedIn optimise every layer for AI workloads 28 Sept
• Staying on its own data centres instead of moving to Azure came down to timing and scale 28 Sept
• A Postgres scaling result from GCP is impressive 25 Sept
• Candidates are surprised by LinkedIn's own data centres, traffic scale and AI work 24 Sept
• Recommends a listen on how pop hits were made 19 Sept
• A cloud-seeding result is striking 25 Aug
• Recommends a book on building safer technology 21 Aug
• The next season of InTheTech is filming 13 Aug
• Agrees early-career people should optimise for opportunity over salary 13 Aug
• Agrees AI changes what great engineers do 1 Aug
• LinkedIn aims to deliver more AI while keeping compute close to flat 30 Jul
• Owning its data centres makes that possible 30 Jul
• An AI accelerator is open to economic-mobility nonprofits 27 Jul
• LLMs can improve ads ranking if made fast and cheap enough for LinkedIn's scale 23 Jul

Posts

Erran Berger28 SeptEngineering and infrastructure

Owning our technology stack end-to-end enables us to optimize every layer - something that becomes increasingly important as we deploy ever-hungrier AI workloads. When we decided to double down on our own data centers and infrastructure rather than continue… more

Owning our technology stack end-to-end enables us to optimize every layer - something that becomes increasingly important as we deploy ever-hungrier AI workloads. When we decided to double down on our own data centers and infrastructure rather than continue migrating to Azure, it really came down to timing and scale. Azure was growing at an incredible rate (and still is!) because customer demand was going through the roof, and LinkedIn was seeing its own skyrocketing growth at the same time. Building fresh for the platform we needed was the right strategic bet for LinkedIn's future. While the internet's ChatGPT moment hadn't happened yet, we already saw companies like Meta reaping the benefits of developing larger and larger AI models, and the leverage from the systems they were building to support them. We bet on building that same kind of foundation ourselves and years later, the payoff is that our engineers can design entire systems end-to-end: intelligently splitting inference workloads across GPUs and CPUs, distilling large models into smaller ones that are very good at solving a specific optimization problem, and going as deep as writing our own GPU kernels with Liger Kernel to squeeze more out of every training run. Owning all of this made our teams more disciplined and more creative at the same time. Thanks to Frederic Lardinois for a great conversation at the WeAreDevelopers Conference - video to come!

88 reactions · 2 comments · – repostsOpen on LinkedIn

Erran Berger25 SeptRepost with commentaryEngineering and infrastructure

This is truly awesome.

Urs Hölzle

A huge first in databases: dynamically scale up a true postgres DB to 1,000 machines and back to 4 cores in two minutes! This is real SQL, not a subset: the… more

A huge first in databases: dynamically scale up a true postgres DB to 1,000 machines and back to 4 cores in two minutes! This is real SQL, not a subset: the full power of relational SQL, hybrid search (vector, full-text, spatial), and indexes, one terabit/sec throughput. Only GCP can do that 👍 More detail here: https://lnkd.in/gcuiD3-u

25 reactions · – comments · – repostsOpen on LinkedIn

Erran Berger24 SeptEngineering and infrastructure

Through 17 years of working at LinkedIn, I think I’ve interviewed thousands of candidates and there are a few things that consistently surprise people about us. From operating our own datacenters to our traffic scale to the innovation we're doing with AI… more

Through 17 years of working at LinkedIn, I think I’ve interviewed thousands of candidates and there are a few things that consistently surprise people about us. From operating our own datacenters to our traffic scale to the innovation we're doing with AI across both consumer and enterprise, it's a pretty remarkable range. I recently sat down with Tim Jurka and Wenjing Zhang to discuss some of the technical challenges we're tackling that aren’t always obvious from the outside. Check it out!

168 reactions · 7 comments · 1 repostOpen on LinkedIn

Erran Berger19 Sept

Clicked this link on a whim this morning and after 5 minutes I was completely sucked in. I always love hearing how people at the top of their craft actually work… especially when the biggest breakthroughs are sometimes totally accidental. If you’ve got an… more

Clicked this link on a whim this morning and after 5 minutes I was completely sucked in. I always love hearing how people at the top of their craft actually work… especially when the biggest breakthroughs are sometimes totally accidental. If you’ve got an hour this weekend and love pop music, have a listen. The stories behind Moves Like Jagger (Maroon 5) and Diamonds (Rhianna) alone are worth it. https://lnkd.in/dawGqg8G

57 reactions · 10 comments · 1 repostOpen on LinkedIn

Erran Berger25 AugRepost with commentary

This is wild.

Augustus Doricko

Rainmaker just produced ~19M gallons of water in Alaska via next-gen cloud seeding over 3 hours of operations. We are the first company to provably produce… more

Rainmaker just produced ~19M gallons of water in Alaska via next-gen cloud seeding over 3 hours of operations. We are the first company to provably produce precipitation in Alaska. As promised, we’ve linked our white paper and relevant data. In the future, Rainmaker will protect and restore glaciers with man-made snowfall. Immediately, this demonstration shows how Rainmaker will add new water to the Colorado River and Great Salt Lake in the coming months. We turned around this analysis and white paper just a couple days after operating. There are many data sources I want to deepen our understanding of the atmosphere and increase our provable production; we’re building the instruments to collect that data in subsequent operations. But at Rainmaker, we won’t tout LOIs, simulations, or lab tests. What matters is physically measuring that you’ve affected the atmosphere; proving that we’ve produced water. We prefer sharing only the tech that has proven to make more water for farms, industries, and ecosystems in need. Expect a regular cadence of these from us as we build the best atmospheric science lab in the world. Feedback is welcome in the interim. Rainmaker Century. https://lnkd.in/gHi8YQSB

72 reactions · 9 comments · – repostsOpen on LinkedIn

Erran Berger25 AugRepostAI and the future of work

Abigail Carlton

AI is reshaping how we work. But are we doing enough for young people to navigate this shift? Young talent is uniquely positioned to shape how AI transforms… more

AI is reshaping how we work. But are we doing enough for young people to navigate this shift? Young talent is uniquely positioned to shape how AI transforms the world of work, if we set them up for success. This means employers, educators, and workforce organizations need to ensure they continue to have access to meaningful early-career opportunities. In our Forbes op-ed, Susan Murray and I share why rethinking career pathways matters for the next generation, and for the organizations and economy they'll help build. https://lnkd.in/gmevusBU

Open on LinkedIn

Erran Berger21 AugRepost with commentaryText cut shortEngineering and infrastructure

Must read alert!

Lea Kissner

I can't believe that our new book, "Building Safer Technology: A Field Guide to Failing Well" is here and in print. This book teaches how to think about… more

I can't believe that our new book, "Building Safer Technology: A Field Guide to Failing Well" is here and in print. This book teaches how to think about building technology in a world where things fail -- security, privacy, trust&safety, AI safety, and more. Because we concentrate on the underlying thinking and skills, this is durable knowledge which you can use even as the details of the technology changes every single day. I don't have a copy yet, so you might be able to get one before I do! Get it from the publisher: https://lnkd.in/gG3G7Bzn Or Amazon: https://lnkd.in/g4r2QH8Y

51 reactions · 5 comments · 1 repostOpen on LinkedIn

Erran Berger13 AugEngineering and infrastructureShows and podcasts

Next “season” of #InTheTech filming today with these two characters. Thanks for being good sports, Wenjing & Tim! Can’t wait to see the final cuts!

279 reactions · 24 comments · 1 repostOpen on LinkedIn

Erran Berger13 AugRepost with commentaryNetworking and careers

This times a million.

Dan Porter

My take for early career people is, don’t optimize on salary, optimize on opportunity. Salary is great. But it can be a false indicator. And you can spend… more

My take for early career people is, don’t optimize on salary, optimize on opportunity. Salary is great. But it can be a false indicator. And you can spend it on cool sh-t or even pay bills, but you can’t use it to invest in your future. If you’re 25 you might have 40 years of work left. You won’t remember the 10k more but you will remember the boss who changed your life or the opportunity that opened your mind or the when you learned what risks to take or not to take. You should make as much money as you want or need. But that’s the short game. You’ll do better in the end if you play the long game. Opportunity should be priced in to your decisions, because it's a mistake to just think about salary. And tbh, it's really not much more complicated than that.

47 reactions · 5 comments · – repostsOpen on LinkedIn

Erran Berger7 AugRepost

Max Simkoff

I don’t post on LinkedIn very often. But this one is personal. AI is about to mint a staggering number of new millionaires and billionaires. In fact,… more

I don’t post on LinkedIn very often. But this one is personal. AI is about to mint a staggering number of new millionaires and billionaires. In fact, according to ChatGPT, The Bay Area alone could see $200–300B of new liquid wealth over the next few years (ironically - most of this coming from people who work for ChatGPT's parent OpenAI). And I keep hearing that these people should reinvent philanthropy. Why? We already know how to do philanthropy. There are incredible organizations filled with people who have spent decades figuring out how to solve hard problems. They don’t need Silicon Valley to build them a new operating system. They need Silicon Valley to write the check. My own deep personal involvement with JVS - Bay Area has made this particularly clear to me. There are remarkable organizations doing profoundly important work—but far too often, the constraint isn't ideas. It’s resources. That’s the argument Lisa Countryman-Quiroz and I make in our new Fortune Op-Ed: https://lnkd.in/e482aYfA So, for all you already (or soon-to-be) AI millionaires (or deca/centa millionaires, or billionaires) - please, pick a capable philanthropic platform that already exists and give (generously) to help here.

Open on LinkedIn

Erran Berger7 AugRepost

Lisa Countryman-Quiroz

AI is creating extraordinary wealth. We have a once-in-a-generation opportunity to ensure it also expands opportunity. In our new Fortune op-ed, Max Simkoff… more

AI is creating extraordinary wealth. We have a once-in-a-generation opportunity to ensure it also expands opportunity. In our new Fortune op-ed, Max Simkoff and I make the case that the next wave of AI philanthropy should invest in the nonprofits already helping people and communities thrive. Across the country, organizations are responding to changing needs every day, building partnerships, innovating, and creating pathways to economic mobility. At JVS - Bay Area, we've seen how quickly the workforce is evolving. We've adapted our training programs, integrated AI skills across our services, and continue to partner with employers to prepare people for quality careers in an AI-enabled economy. Imagine what organizations like ours could do with greater investment. I'm especially grateful to Max for his partnership on this piece. As Founder and CEO of Doma Technology LLC and a dedicated JVS board member, he shares a deep belief that innovation and equity must go hand in hand. It was a privilege to write this together. AI has the potential to be transformational, but only if we invest in the people and organizations building a future where opportunity is within everyone's reach. Read the full piece here: https://lnkd.in/gTBw7Yye #AI #WorkforceDevelopment #EconomicMobility #Philanthropy #FutureOfWork

Open on LinkedIn

Erran Berger7 AugRepostEngineering and infrastructure

Balaji Krishnapuram

#KDD2026 is a wonderful opportunity to meet researchers and practitioners to discuss ideas that are shaping the future of AI. If you're joining us at KDD 2026,… more

#KDD2026 is a wonderful opportunity to meet researchers and practitioners to discuss ideas that are shaping the future of AI. If you're joining us at KDD 2026, here are a few LinkedIn talks I’d like to share: Generative Recommendation at LinkedIn: Foundation Models, Semantic IDs, and the Cold-Start Problem Bo Long Tuesday, August 11 | 12:30-1:00 PM Location: Exhibit Hall Shifting the Unit of Safety: From Model to System in the Generative and Agentic Era Sakshi Jain Wednesday, August 12 | 10:00-11:00 AM Location: Tamna Hall Scaling Abuse Prevention for the Next Wave of AI at LinkedIn Daniel Olmedilla Sunday, August 9 | 11:05-11:30 AM Location: 303A Stop by one of these sessions or come visit us at LinkedIn Booth #1 & #2. Looking forward to the conversations, exchanging ideas with the community, and seeing many of you in Jeju! #KDD2026 #LinkedInEng #AIResearch #LLM #DataScience

Open on LinkedIn

Erran Berger6 AugRepostAI and the future of work

Inspire Leadership Network

Talent isn’t separate from AI strategy. It is AI strategy. Erran Berger and Susan Murray challenged #Converge26 attendees to reject the false choice between… more

Talent isn’t separate from AI strategy. It is AI strategy. Erran Berger and Susan Murray challenged #Converge26 attendees to reject the false choice between becoming AI-led and employee-first. “The companies that are able to transform their talent and technology bases are going to be the ones that really thrive in the age of AI.“

Open on LinkedIn

Erran Berger5 AugRepost

Lisa Countryman-Quiroz

This The New York Times DealBook piece asks a question the nonprofit sector has been sitting with for awhile: When AI IPO wealth starts flowing to… more

This The New York Times DealBook piece asks a question the nonprofit sector has been sitting with for awhile: When AI IPO wealth starts flowing to philanthropy, where will it go? A couple months ago, there was a Substack piece (you likely know the one) that tried to tackle this question and caused quite a stir in nonprofit circles. It argued that the nonprofit sector is "orders of magnitude" away from being able to absorb incoming AI philanthropic capital — and that we need new "philanthropic startups" with "tech-caliber talent" to get the job done. We have to bridge this fundamental misunderstanding in order to deploy AI wealth for the greatest good. And I’d like to start by inviting tech leaders to learn more about my sector before they decide to build from scratch. Here’s what you need to know: The nonprofit sector isn't a blank slate waiting for tech to arrive with better ideas or more efficient execution. It's the infrastructure behind some of the most consequential progress in modern history — from the expansion of public education to the response to the AIDS crisis to the environmental movement that gave us the Clean Air and Clean Water Acts. Nonprofits (almost 2 million of them in the U.S. alone!) are run by talented, tenacious people who have spent careers learning exactly what it takes to solve enormous societal challenges. They know how to build trust in communities that have been let down before, how to navigate the gap between policy and people's actual lives, and, for better or worse, how to do more with less year after year. At JVS - Bay Area, we've been helping workers access stable, well-paid jobs for 50 years. When AI started reshaping employment at high speed, we didn't wait for someone to build a new organization to address it. We sunsetted programs that were no longer serving jobseekers, built new ones in sectors more resistant to automation, and integrated AI skills across our training. Despite all the headwinds, our graduates are generally able to secure jobs within less than a month of completing our programs. That's what nonprofits do — we adapt. Because adapting and innovating is how we change peoples’ lives for the better. The nonprofit sector has been waiting a long time for partners with the resources to match its ambition. If you're thinking seriously about where your philanthropic dollars should go, I hope you'll start by getting curious about what already exists. There's a lot here worth knowing. https://lnkd.in/gZDHRRGJ #Philanthropy #NonprofitSector #AI

Open on LinkedIn

Erran Berger1 AugRepost with commentaryAI and the future of workEngineering and infrastructure

Spot on Alborz.

Alborz Geramifard

AI isn't eliminating engineering. It's changing what great engineers do. A conversation with Ashwin Machanavajjhala and Erran Berger inspired me to think… more

AI isn't eliminating engineering. It's changing what great engineers do. A conversation with Ashwin Machanavajjhala and Erran Berger inspired me to think about this through the lens of one of computing's biggest transitions: from assembly to Python. I believe we're living through an equally important shift today. Here's my perspective.

28 reactions · 1 comment · 2 repostsOpen on LinkedIn

Erran Berger30 JulEngineering and infrastructure

As AI moves from experimentation to production, discipline is going to matter just as much as ambition. At LinkedIn we saw a while ago that the cost trends were clear, and that the future of our products would be powered by increasingly compute-expensive AI… more

As AI moves from experimentation to production, discipline is going to matter just as much as ambition. At LinkedIn we saw a while ago that the cost trends were clear, and that the future of our products would be powered by increasingly compute-expensive AI workloads that are more intelligent, more capable, and more useful. But those workloads have to be ROI positive, and that’s not something you can just wake up and do overnight. So we set ourselves a challenge that sounds counterintuitive in today’s environment: how do we deliver increasingly sophisticated AI experiences while keeping our compute footprint as close to flat as possible? This is where owning our own data centers and infrastructure as an applied ML company really pays off, because we can put in the instrumentation and make efficiency a long term investment up and down the stack rather than a one-time push. None of it came from a single breakthrough - it’s the outcome of hundreds of improvements, from optimizing GPU utilization and workload allocation to distilling larger models into smaller and more efficient ones to rethinking how work is distributed across training, inference, storage, and systems design. The results of that intentional push are starting to play out in terms of better experiences for our members and customers while giving our engineers the flexibility to keep innovating. Raghu and I recently sat down with Paresh Dave at WIRED to talk through all of this, check it out! https://lnkd.in/guYj674d

1,007 reactions · 41 comments · 26 repostsOpen on LinkedIn

Erran Berger27 JulRepost with commentary

Huge opportunity for any nonprofits in the economic mobility or workforce development space. Check it out!

Lisa Countryman-Quiroz

I had such a great time at Jobs for the Future (JFF) #Horizons last week presenting with my colleagues Caitlyn Brazill, Meghan Cressman, Scott Gullick and… more

I had such a great time at Jobs for the Future (JFF) #Horizons last week presenting with my colleagues Caitlyn Brazill, Meghan Cressman, Scott Gullick and Arathi Ravier and Leading Educators on our accomplishments from the first cohort of of the Economic Mobility AI Accelerator! If you work in the economic mobility space and are ready to move beyond experimentation, come learn with us! The second Economic Mobility AI Accelerator is open for applications now. There is truly no better way to learn than by doing!

14 reactions · 2 comments · – repostsOpen on LinkedIn

Erran Berger23 JulEngineering and infrastructure

The most interesting AI stories aren’t about the model, they’re about constraints. At LinkedIn, we had a hypotheses that LLMs can be applied to traditional ranking and recommendations problems by better understanding what members find relevant. But using… more

The most interesting AI stories aren’t about the model, they’re about constraints. At LinkedIn, we had a hypotheses that LLMs can be applied to traditional ranking and recommendations problems by better understanding what members find relevant. But using them for a domain like ads is far from straightforward. Decisions need to happen at enormous scale and in a fraction of a second, which means the usual approach quickly becomes too slow and too expensive. That’s the engineering challenge Xīan Xíng Zhāng, Distinguished Software Engineer for Ads AI, and his team took on. They designed a way to use LLMs to better match what members are looking for with what advertisers are offering, while maintaining the efficiency and performance needed to operate at the scale of LinkedIn.

131 reactions · 1 comment · 5 repostsOpen on LinkedIn