← Journal · · Personal
Rebrand and plans for the year
Hey everyone! I've finally recovered after the holidays and I'm ready to share some news about this channel with you 🔥
1. I found my area of interest: ML infrastructure 💪. I created this channel a little less than a year ago. At first I thought I'd write about DevOps in plain language, then the content drifted toward MLOps and my articles/talks, and I did a rebrand, trying to ironically play with the mainstream trend of creating your own channels. But the main reason behind changing the channel's names and concepts is my search for my own niche and area of interest. We all grow and change over time, so I don't think it's a bad thing to occasionally "change lanes" and shift direction.
Reflecting on my current content and areas of interest, I can say that in the near future I want to go deeper into the world of ML infrastructure. My main research is aimed exactly at this topic, I already have expertise in building ML platforms, inference, and working with GPUs in the cloud. And going forward I want to expand my expertise in this area and share it with you, of course. So I decided to rename the channel again, and also add more of my personality to it :)
2. Plans for the year, research-wise This year I'm planning to:
- Study Dynamic Resource Allocation in k8s (want to wait for version 1.33) and show you new ways to manage GPU resources in k8s.
- Dig deeper into autoscaling optimization, take a closer look at lazy loading and snapshotters, the zstd format, and p2p storage.
- Keep building expertise in Triton Inference Server, study inference optimization methods. How to make GPUs more accessible with minimal quality loss for the model.
- Continue writing articles and speaking at conferences.
3. You can always reach out to me for help For several months now I've been getting questions and requests about working with ML infrastructure. For small questions I'm happy to help (for example, setting up the GPU operator, or questions about LLM inference).
But if you have a complex problem, you can also reach out to me for a consultation, for example through this platform, on topics like:
- Building infrastructure for ML tasks
- Choosing infrastructure configuration for LLMs
- Help developing inference services
- A career development track for an ML infrastructure specialist
I'm really glad I've found a direction that I like a lot right now, and that it's interesting to you too :) Stay tuned and expect more content, see you soon! 😊