Michael Burkov
@xmikebur
Followers
38
Following
87
Media
7
Statuses
130
GTM @tractoai. Simple & powerful runtime for AI and Data teams. Prev, EIR @ Primary VC, Google engineer
New York, NY
Joined September 2023
RL workloads are bursty by nature. Plan capacity to avoid over/under‑provisioning, and make autoscaling a first‑class feature.
0
0
0
Batch processing is everywhere. Classic map reduce is a popular approach for working with AI data. Map operation gives you flexibility to run custom code on your data to prepare datasets for training and RL. Cursor team talked a lot about it.
1
0
0
Reliability across nodes matters most. In distributed jobs, hardware fails (not if but when). Building a system that restarts failed jobs in the multi node environment is a complicated engineering task. I heard this from many people.
0
0
0
AI native compute is here. In the context of ML, we are used to data processing running on CPUs, model training and serving on GPUs. This now is consolidating around CPU/GPU compute hardware capable of processing various workloads. Ray leads the way here.
0
0
0
Came back from #raysummit by @anyscalecompute . I spoke w/ and learned from AI operators at Notion, Uber, xAI and the like. AI infra trends affecting engineers working in the space:
3
0
0
As foundational models converge in capability, the true edge for AI products comes down to 2 things: 1. How fast you can iterate. 2. How well you can use data to improve model behavior. Learn from Cursor https://t.co/zrAOXNxVic
0
0
0
Helping my friend Garrett whose new venture makes pet care more affordable & helps save pets from euthanasia. Sign up in 30 sec:
subscribepage.io
Be part of the future of pet healthcare. Sign up to help support the animals you love — and help make lifesaving care accessible to all.
0
0
0
Monetize yourself !
Today, we’re opening a new chapter in knowledge work. 🚀 We’re finally ready to share MyClone with the world. After having 100+ conversations with top knowledge professionals (consultants, coaches, advisors) like Balki Kodarapu, @peterahn , Riley, Mor Milo, Walter Roth, Mariya
1
0
3
Spot on.
@nebiusai is way more than just compute, we’re building a full developer platform to power AI products at scale. A robust data platform is at the core of this vision, enabling data intensive workloads for AI. https://t.co/fwCoPle2P6 New article on how we make this possible.
0
0
0
AI is pushing traditional data stacks to their breaking point. ⚠️ In his new blog, Maxim Akhmedov, Head of TractoAI at Nebius, explains why unified, scalable infra is essential for multimodal data + inference at AI scale and why DIY point solutions are too costly to manage.
2
19
138
I’ll start to ask for feedback on different products we do. First one The coolest research effort at @nebiusai: SWE-rebench is the best benchmark out there for coding agents. Drop a comment to @agolubev13 and the team! Pls share 🙌
What are the key evaluation directions you'd like to see for LLMs as SWE-agents? In SWE-rebench, we measure how well models solve GitHub issues. We're considering extensions, primarily measuring how well models write tests for these fixes, and extending the benchmark to other
2
9
96
Marc with John and Jordi. Yippee
IN NEWS: Nebius lands a $17.4B partnership with Microsoft to provide high-performance AI infrastructure. The CRO @marcboroditsky told us: "We believe the real opportunity is delivering the AWS equivalent for AI.” “The $100B+ business opportunity is servicing the
0
0
0
we heard there are some challenges with getting OpenAI oss models running with vllm out of the box. Solved it
If you itching to experiment with gpt oss 120b using vllm here's a working notebook on our GPU sandbox. Run, experiment, share for free. https://t.co/ehHLY56XBD
0
0
0
If you itching to experiment with gpt oss 120b using vllm here's a working notebook on our GPU sandbox. Run, experiment, share for free. https://t.co/ehHLY56XBD
0
1
1
Last time I saw something like this was in the Inception movie
🔘 Promptable world events Beyond navigation, users can insert text prompts to alter the world in real-time - like changing the weather ⛅ or introducing new characters 👤 This unlocks a new level of dynamic interaction.
0
0
0
Will everyday consumer AI be subsidized by ads or not?
Mark Cuban is promoting a policy that would almost certainly make the best AI inaccessible for low income Americans and lead to greater wealth inequality over time
0
0
0
Names of new models look complicated. What does it even mean? True it’s the same as “Apple M3 258 Gb Ram , etc” Let make it less intimating
what an intense month it has been. free models we received in the last few weeks: - kimi k2 1t 32b - qwen3 235b a22b non-thinking - qwen3 coder 480b a35b - new magistral 24b from mistral - qwen3 235b a22b reasoning - step3 321b a38b with new mfa attention - smollm3 3b - intern
0
0
1
Demos, mostly Agentic Apps, starting at 1pm today @theaifurnace . Excited to see what's possible today vs a year ago. Big thanks to all the teams taking time to demo their products in front of the audience
0
0
0