Damian Bogunowicz

[Damian Bogunowicz @dtransposed]

My recent blog post about the lessons learned while working fully remotely:
https://t.co/zZGaXUPQTu
has become the number one trending submission on Hacker News today!

[Damian Bogunowicz @dtransposed]

I'll be attending @icmlconf 2026 from July 18–24.
If you'll be around, I'd love to grab a coffee and meet both new and familiar faces.
See you in Seoul! ☕️

[Damian Bogunowicz @dtransposed]

@AkrShakil @nv_pavlichenko This is very important, let’s share it with the creators of Qwen, GLM, Kimi, MiMo or MiniMax.

[Damian Bogunowicz @dtransposed]

@javi_22_dev @nv_pavlichenko We’ve been following this domain for a while in JetBrains. I am personally convinced that these models will be soon ripe for production. We are cooking internally, stay tuned.

[Damian Bogunowicz @dtransposed]

@nmb_four @nv_pavlichenko It depends on how you deploy it, so it’s pointless to mention it in a technical report.

[Damian Bogunowicz @dtransposed]

My DMs are open to excellent engineers who want to build AI coding features on top of the new Mellum model.

Why join? At JetBrains AI, we own the full AI product lifecycle - including distribution, which is rare:
1. We build foundation models: pretraining, post-training, and RL
2. We build coding agents on top of them
3. We build the inference stack to serve AI features efficiently
4. We deploy them to hundreds of thousands of users

[Damian Bogunowicz @dtransposed]

Excellent work @nv_pavlichenko and the team.

[vLLM @vllm_project]

🎉 Congrats to @JetBrains on Mellum2-12B-A2.5B-Thinking, an open-source 12B MoE that activates just 2.5B params, handling both natural language and code with a 128K context.

Mellum2 runs natively in vLLM from day 0, with reasoning parser and tool calling for agentic workflows.

🔗 https://t.co/72E6HDOHNf

[JetBrains @jetbrains]

Mellum started with code completion.
Mellum2 is built for more – handling both natural language and code.
A 12B-parameter open-source LLM for routing, RAG, and sub-agents, optimized for ultra-low-latency inference.
Now on @huggingface.
Learn more: https://t.co/28sG8Ql52L

[Damian Bogunowicz @dtransposed]

Excellent work @nv_pavlichenko and the team.

[Nikita Pavlichenko @nv_pavlichenko]

Today we're releasing Mellum2: our first "serious" LLM.

This is a 12B A2.5B MoE LLM pre-trained on ~11T tokens and post-trained with RLVR.
I'm proud to be leading the team that was working on it for the last 6 months.

We release base/SFT/RL checkpoints along with a tech report

[JetBrains @jetbrains]

We did it again.

Something new for the open-source LLM space is coming to @huggingface soon.

Stay tuned.

[Damian Bogunowicz @dtransposed]

100%

[doomer @uncledoomer]

everybody i know who has kids says that japan is the best place on earth to vacation when you have a baby

[Elon Musk @elonmusk]

Europe will love Tesla self-driving!

Due to the extreme regulatory burden of the EU, which in general stifles innovation in Europe, Tesla owners there have been stuck with basic lane-following.

[Damian Bogunowicz @dtransposed]

@jetbrains at @vllm_project Warsaw meetup

[Red Hat AI @RedHat_AI]

Video 1: vLLM Project Update + @JetBrains + NVIDIA Flex Tensor

@mgoin_ walked through vLLM updates: async scheduling, hybrid KV cache, and disaggregated prefill/decode for multi-node deployments.

@JetBrains runs vLLM in production with FP8 + speculative decoding. Live demo on DGX Spark: 4-bit quantization cut latency from 20s to 1s.

NVIDIA Flex Tensor: automatic layer offloading for models that exceed single GPU memory, with near-zero overhead.

[Damian Bogunowicz @dtransposed]

If you missed it, the Warsaw edition of @vllm_project meet-up with @jetbrains @RedHat_AI and @nvidia was recorded! See all the sessions below:

[Red Hat AI @RedHat_AI]

Warsaw @vllm_project meetup recordings are live 🇵🇱

Video 1: vLLM roadmap, JetBrains AI in IDEs, and NVIDIA Flex Tensor

Video 2: vLLM Omni for multimodal output and @_llm_d_ for distributed inference on Kubernetes

5 sessions. All technical. Thread below 👇

[Damian Bogunowicz @dtransposed]

Timeless principles.

[Jaynit @jaynitx]

In 2019, MIT professor Patrick Winston gave a legendary 1-hour lecture called “How to Speak.”

It has 18M+ views for a reason.

His frameworks:
• Your ideas are like your children
• The 5-minute rule for job talks
• Why jokes fail at the start

15 lessons on communication:

[Damian Bogunowicz @dtransposed]

@mgoin_ 🤢

[Damian Bogunowicz @dtransposed]

Oh, the good old days when one could stick with the same IDE for several years. Now I have Air, c-mux, and cursor/glass installed. I just uninstalled Kiro, but a friend from Google inspired me to try Antigravity. I can’t wait for this space to finally solidify so I can feel less FOMO.

[Damian Bogunowicz @dtransposed]

@cursor_ai after a month in Poland

[Damian Bogunowicz @dtransposed]

GTC is packed holy moly