Alexander Nowak
@alenowak.bsky.social
#python Software Engineer 🐍
#googlecloud Architect
#machinelearning Practitioner
#googlecloud Architect
#machinelearning Practitioner
Reposted by Alexander Nowak
i’ve been experimenting with Nano Banana here and i feel like it’s a brand new medium — tweet + infographic
curious what others think, but data-wise it seems like people appreciate it
ime you’re not guaranteed to get good results out of NB, i still need to understand *before* requesting an image
curious what others think, but data-wise it seems like people appreciate it
ime you’re not guaranteed to get good results out of NB, i still need to understand *before* requesting an image
DSA: DeepSeek Sparse Attention
DeepSeek 3.2 & 3.2-Speciale are ridiculously cheap because of DSA
LLMs aren’t quadratic anymore
They trained an additional “model” that does acts as a “pre-attention”, selecting only the portions that are probably relevant
DeepSeek 3.2 & 3.2-Speciale are ridiculously cheap because of DSA
LLMs aren’t quadratic anymore
They trained an additional “model” that does acts as a “pre-attention”, selecting only the portions that are probably relevant
December 2, 2025 at 4:02 PM
i’ve been experimenting with Nano Banana here and i feel like it’s a brand new medium — tweet + infographic
curious what others think, but data-wise it seems like people appreciate it
ime you’re not guaranteed to get good results out of NB, i still need to understand *before* requesting an image
curious what others think, but data-wise it seems like people appreciate it
ime you’re not guaranteed to get good results out of NB, i still need to understand *before* requesting an image
Reposted by Alexander Nowak
How many releases do you like to have in a day?
Here's more: 0.123.3, 0.123.4 🥞
Granular releases, so you can get the latest, but if something doesn't work for you for some reason, you can still get some of the fixes in the previous release. ☕️
Here's more: 0.123.3, 0.123.4 🥞
Granular releases, so you can get the latest, but if something doesn't work for you for some reason, you can still get some of the fixes in the previous release. ☕️
December 2, 2025 at 4:11 PM
How many releases do you like to have in a day?
Here's more: 0.123.3, 0.123.4 🥞
Granular releases, so you can get the latest, but if something doesn't work for you for some reason, you can still get some of the fixes in the previous release. ☕️
Here's more: 0.123.3, 0.123.4 🥞
Granular releases, so you can get the latest, but if something doesn't work for you for some reason, you can still get some of the fixes in the previous release. ☕️
Reposted by Alexander Nowak
Just released! 🚀🐍
Python 3.14.1
Waiting for the .1 to upgrade? This one's especially for you!
🥧 Deferred type annotation evaluation!
🥧 T-strings!
🥧 Zstandard!
🥧 Syntax highlighting in the REPL!
🥧 And much more!
discuss.python.org/t/python-3-1...
#Python #CPython #Python314 #release
Python 3.14.1
Waiting for the .1 to upgrade? This one's especially for you!
🥧 Deferred type annotation evaluation!
🥧 T-strings!
🥧 Zstandard!
🥧 Syntax highlighting in the REPL!
🥧 And much more!
discuss.python.org/t/python-3-1...
#Python #CPython #Python314 #release
December 2, 2025 at 4:58 PM
Just released! 🚀🐍
Python 3.14.1
Waiting for the .1 to upgrade? This one's especially for you!
🥧 Deferred type annotation evaluation!
🥧 T-strings!
🥧 Zstandard!
🥧 Syntax highlighting in the REPL!
🥧 And much more!
discuss.python.org/t/python-3-1...
#Python #CPython #Python314 #release
Python 3.14.1
Waiting for the .1 to upgrade? This one's especially for you!
🥧 Deferred type annotation evaluation!
🥧 T-strings!
🥧 Zstandard!
🥧 Syntax highlighting in the REPL!
🥧 And much more!
discuss.python.org/t/python-3-1...
#Python #CPython #Python314 #release
Reposted by Alexander Nowak
NEW VERTEX AI ENGINEERING BLOG: Scaling EAGLE-3 on Vertex AI with SGLang
We've launched an engineering blog to document our Vertex AI research. Our first post, in collaboration withSGLang, shares insights on implementing EAGLE-3 at scale.
Blog and code in the 🧵
We've launched an engineering blog to document our Vertex AI research. Our first post, in collaboration withSGLang, shares insights on implementing EAGLE-3 at scale.
Blog and code in the 🧵
December 2, 2025 at 2:30 PM
NEW VERTEX AI ENGINEERING BLOG: Scaling EAGLE-3 on Vertex AI with SGLang
We've launched an engineering blog to document our Vertex AI research. Our first post, in collaboration withSGLang, shares insights on implementing EAGLE-3 at scale.
Blog and code in the 🧵
We've launched an engineering blog to document our Vertex AI research. Our first post, in collaboration withSGLang, shares insights on implementing EAGLE-3 at scale.
Blog and code in the 🧵
Reposted by Alexander Nowak
Nano Banana Pro, released this morning, is clearly the best image generation model. Superb instruction following, plus it can generate full infographics (with correct spelling and properly rendered text!) from a short prompt based on running extra searches simonwillison.net/2025/Nov/20/...
Nano Banana Pro aka gemini-3-pro-image-preview is the best available image generation model
Hot on the heels of Tuesday’s Gemini 3 Pro release, today it’s Nano Banana Pro, also known as Gemini 3 Pro Image. I’ve had a few days of preview access …
simonwillison.net
November 20, 2025 at 4:34 PM
Nano Banana Pro, released this morning, is clearly the best image generation model. Superb instruction following, plus it can generate full infographics (with correct spelling and properly rendered text!) from a short prompt based on running extra searches simonwillison.net/2025/Nov/20/...
Reposted by Alexander Nowak
Nano Banana Pro
A reasoning model image generator!
- multi turn
- google search grounding
- interleaved image & text
console.cloud.google.com/vertex-ai/pu...
A reasoning model image generator!
- multi turn
- google search grounding
- interleaved image & text
console.cloud.google.com/vertex-ai/pu...
You are using an unsupported browser. If you see unexpected behavior, upgrade to a supported browser. Learn more.
console.cloud.google.com
November 20, 2025 at 3:42 PM
Nano Banana Pro
A reasoning model image generator!
- multi turn
- google search grounding
- interleaved image & text
console.cloud.google.com/vertex-ai/pu...
A reasoning model image generator!
- multi turn
- google search grounding
- interleaved image & text
console.cloud.google.com/vertex-ai/pu...
Reposted by Alexander Nowak
I’m really excited about our release of Gemini 3 today, the result of hard work by many, many people in the Gemini team and all across Google! 🎊
blog.google/products/gem...
Gemini 3 performs quite well on a wide range of benchmarks.
blog.google/products/gem...
Gemini 3 performs quite well on a wide range of benchmarks.
November 19, 2025 at 2:53 AM
I’m really excited about our release of Gemini 3 today, the result of hard work by many, many people in the Gemini team and all across Google! 🎊
blog.google/products/gem...
Gemini 3 performs quite well on a wide range of benchmarks.
blog.google/products/gem...
Gemini 3 performs quite well on a wide range of benchmarks.
Reposted by Alexander Nowak
Google Antigravity: an agentic-first software IDE
a ground-up redesign of software development with agents at the center of
antigravity.google
a ground-up redesign of software development with agents at the center of
antigravity.google
November 18, 2025 at 5:09 PM
Google Antigravity: an agentic-first software IDE
a ground-up redesign of software development with agents at the center of
antigravity.google
a ground-up redesign of software development with agents at the center of
antigravity.google
Reposted by Alexander Nowak
A new era of intelligence with Gemini 3
YouTube video by Google DeepMind
www.youtube.com
November 18, 2025 at 4:45 PM
Reposted by Alexander Nowak
Google's most advanced model, Gemini 3, is now live!
📈 1487 Elo on WebDev Arena, 76.2% on SWE-bench Verified
🛠️ Try out Google Antigravity: A new agentic IDE with direct access to the terminal, editor, and browser to build and validate code.
blog.google/products/gem...
📈 1487 Elo on WebDev Arena, 76.2% on SWE-bench Verified
🛠️ Try out Google Antigravity: A new agentic IDE with direct access to the terminal, editor, and browser to build and validate code.
blog.google/products/gem...
A new era of intelligence with Gemini 3
Today we’re releasing Gemini 3 – our most intelligent model that helps you bring any idea to life.
blog.google
November 18, 2025 at 4:16 PM
Google's most advanced model, Gemini 3, is now live!
📈 1487 Elo on WebDev Arena, 76.2% on SWE-bench Verified
🛠️ Try out Google Antigravity: A new agentic IDE with direct access to the terminal, editor, and browser to build and validate code.
blog.google/products/gem...
📈 1487 Elo on WebDev Arena, 76.2% on SWE-bench Verified
🛠️ Try out Google Antigravity: A new agentic IDE with direct access to the terminal, editor, and browser to build and validate code.
blog.google/products/gem...
Reposted by Alexander Nowak
Supervised Fine-Tuning (SFT) adapts pre-trained models with labeled data but often misses aligning with user preferences.
Vertex AI now supports preference tuning (DPO) for Gemini 2.5 Flash and Flash-Lite, allowing you to use response pairs to adjust user preferences.
Code and docs in the 🧵
Vertex AI now supports preference tuning (DPO) for Gemini 2.5 Flash and Flash-Lite, allowing you to use response pairs to adjust user preferences.
Code and docs in the 🧵
November 14, 2025 at 6:00 AM
Supervised Fine-Tuning (SFT) adapts pre-trained models with labeled data but often misses aligning with user preferences.
Vertex AI now supports preference tuning (DPO) for Gemini 2.5 Flash and Flash-Lite, allowing you to use response pairs to adjust user preferences.
Code and docs in the 🧵
Vertex AI now supports preference tuning (DPO) for Gemini 2.5 Flash and Flash-Lite, allowing you to use response pairs to adjust user preferences.
Code and docs in the 🧵
Reposted by Alexander Nowak
pytest 9.0 just released!
Notable changes:
* subtests support
* Native TOML configuration
* Progress output in the terminal taskbar
And others. Check the changelog: docs.pytest.org/en/stable/ch...
Notable changes:
* subtests support
* Native TOML configuration
* Progress output in the terminal taskbar
And others. Check the changelog: docs.pytest.org/en/stable/ch...
Changelog - pytest documentation
docs.pytest.org
November 8, 2025 at 8:07 PM
pytest 9.0 just released!
Notable changes:
* subtests support
* Native TOML configuration
* Progress output in the terminal taskbar
And others. Check the changelog: docs.pytest.org/en/stable/ch...
Notable changes:
* subtests support
* Native TOML configuration
* Progress output in the terminal taskbar
And others. Check the changelog: docs.pytest.org/en/stable/ch...
Reposted by Alexander Nowak
Introducing the File Search Tool in the Gemini API, our hosted RAG solution with free storage and free query time embeddings 💾
We are super excited about this new approach and think it will dramatically simplify the path to context aware AI systems, more details in 🧵
We are super excited about this new approach and think it will dramatically simplify the path to context aware AI systems, more details in 🧵
November 6, 2025 at 6:42 PM
Introducing the File Search Tool in the Gemini API, our hosted RAG solution with free storage and free query time embeddings 💾
We are super excited about this new approach and think it will dramatically simplify the path to context aware AI systems, more details in 🧵
We are super excited about this new approach and think it will dramatically simplify the path to context aware AI systems, more details in 🧵
Reposted by Alexander Nowak
We just shipped some nice Gemini API updates for developers using Structured Outputs. The API now supports:
- $ ref for recursive schemas
- anyOf union types
- min + max numerical constraints
- null types
- property ordering adherence
And much more!
- $ ref for recursive schemas
- anyOf union types
- min + max numerical constraints
- null types
- property ordering adherence
And much more!
November 5, 2025 at 6:06 PM
We just shipped some nice Gemini API updates for developers using Structured Outputs. The API now supports:
- $ ref for recursive schemas
- anyOf union types
- min + max numerical constraints
- null types
- property ordering adherence
And much more!
- $ ref for recursive schemas
- anyOf union types
- min + max numerical constraints
- null types
- property ordering adherence
And much more!
Reposted by Alexander Nowak
The Smol Training Playbook: The Secrets to Building World-Class LLMs
A practical journey through the challenges, decisions, and messy reality behind training state-of-the-art language models.
huggingface.co/spaces/Huggi...
A practical journey through the challenges, decisions, and messy reality behind training state-of-the-art language models.
huggingface.co/spaces/Huggi...
October 31, 2025 at 2:06 AM
The Smol Training Playbook: The Secrets to Building World-Class LLMs
A practical journey through the challenges, decisions, and messy reality behind training state-of-the-art language models.
huggingface.co/spaces/Huggi...
A practical journey through the challenges, decisions, and messy reality behind training state-of-the-art language models.
huggingface.co/spaces/Huggi...
Reposted by Alexander Nowak
A short talk on the main architecture components of LLMs this year + a look beyond the transformer architecture: www.youtube.com/watch?v=lONy...
October 27, 2025 at 3:45 PM
A short talk on the main architecture components of LLMs this year + a look beyond the transformer architecture: www.youtube.com/watch?v=lONy...
Reposted by Alexander Nowak
🤗 Sentence Transformers is joining @hf.co! 🤗
This formalizes the existing maintenance structure, as I've personally led the project for the past two years on behalf of Hugging Face. I'm super excited about the transfer!
Details in 🧵
This formalizes the existing maintenance structure, as I've personally led the project for the past two years on behalf of Hugging Face. I'm super excited about the transfer!
Details in 🧵
October 22, 2025 at 1:04 PM
🤗 Sentence Transformers is joining @hf.co! 🤗
This formalizes the existing maintenance structure, as I've personally led the project for the past two years on behalf of Hugging Face. I'm super excited about the transfer!
Details in 🧵
This formalizes the existing maintenance structure, as I've personally led the project for the past two years on behalf of Hugging Face. I'm super excited about the transfer!
Details in 🧵
Reposted by Alexander Nowak
Just released! 🚀
Please install and enjoy Python 3.14! 🥧
discuss.python.org/t/python-3-1...
#Python #Python314 #release
Please install and enjoy Python 3.14! 🥧
discuss.python.org/t/python-3-1...
#Python #Python314 #release
October 7, 2025 at 2:29 PM
Just released! 🚀
Please install and enjoy Python 3.14! 🥧
discuss.python.org/t/python-3-1...
#Python #Python314 #release
Please install and enjoy Python 3.14! 🥧
discuss.python.org/t/python-3-1...
#Python #Python314 #release
Reposted by Alexander Nowak
JetBrains × Zed: Open Interoperability for AI Coding Agents in Your IDE
blog.jetbrains.com/ai/2025/10/j...
blog.jetbrains.com/ai/2025/10/j...
JetBrains × Zed: Open Interoperability for AI Coding Agents in Your IDE | The JetBrains Blog
JetBrains is collaborating with Zed on the Agent Client Protocol (ACP) – an open protocol that lets AI coding agents work inside editors.
blog.jetbrains.com
October 6, 2025 at 6:39 PM
JetBrains × Zed: Open Interoperability for AI Coding Agents in Your IDE
blog.jetbrains.com/ai/2025/10/j...
blog.jetbrains.com/ai/2025/10/j...
Reposted by Alexander Nowak
PyCharm turns 15! 🎉 Ever wonder about the story behind PyCharm?
Hear directly from the creators as they share the challenges, ideas, and inspiration behind the name.
▶️ Watch the story by @cultrepo.bsky.social now: jb.gg/kz69th
PyCharm: 15 years with Python. And the journey continues.
Hear directly from the creators as they share the challenges, ideas, and inspiration behind the name.
▶️ Watch the story by @cultrepo.bsky.social now: jb.gg/kz69th
PyCharm: 15 years with Python. And the journey continues.
September 25, 2025 at 12:16 PM
PyCharm turns 15! 🎉 Ever wonder about the story behind PyCharm?
Hear directly from the creators as they share the challenges, ideas, and inspiration behind the name.
▶️ Watch the story by @cultrepo.bsky.social now: jb.gg/kz69th
PyCharm: 15 years with Python. And the journey continues.
Hear directly from the creators as they share the challenges, ideas, and inspiration behind the name.
▶️ Watch the story by @cultrepo.bsky.social now: jb.gg/kz69th
PyCharm: 15 years with Python. And the journey continues.
Reposted by Alexander Nowak
ADK v1.15.0 JUST DROPPED with context caching, new observability feature, and Google Maps integration!
Building AI agents with ADK has been significantly upgraded to reduce latency and costs while grounding with Google Maps data and more.
Release notes:
> github.com/google/adk-...
Building AI agents with ADK has been significantly upgraded to reduce latency and costs while grounding with Google Maps data and more.
Release notes:
> github.com/google/adk-...
September 26, 2025 at 6:17 AM
ADK v1.15.0 JUST DROPPED with context caching, new observability feature, and Google Maps integration!
Building AI agents with ADK has been significantly upgraded to reduce latency and costs while grounding with Google Maps data and more.
Release notes:
> github.com/google/adk-...
Building AI agents with ADK has been significantly upgraded to reduce latency and costs while grounding with Google Maps data and more.
Release notes:
> github.com/google/adk-...
Reposted by Alexander Nowak
✅ SUPPORTED: One-Line ADK CLI Deployment to Vertex AI Agent Engine!
Many of you asked, and we've simplified the process: transitioning from an ADK agent to a scalable agent on Vertex AI is now just a single CLI command.
Check out the new docs to try!
👉 Docs: google.github.io/adk-docs/de...
Many of you asked, and we've simplified the process: transitioning from an ADK agent to a scalable agent on Vertex AI is now just a single CLI command.
Check out the new docs to try!
👉 Docs: google.github.io/adk-docs/de...
September 18, 2025 at 4:14 PM
✅ SUPPORTED: One-Line ADK CLI Deployment to Vertex AI Agent Engine!
Many of you asked, and we've simplified the process: transitioning from an ADK agent to a scalable agent on Vertex AI is now just a single CLI command.
Check out the new docs to try!
👉 Docs: google.github.io/adk-docs/de...
Many of you asked, and we've simplified the process: transitioning from an ADK agent to a scalable agent on Vertex AI is now just a single CLI command.
Check out the new docs to try!
👉 Docs: google.github.io/adk-docs/de...
Reposted by Alexander Nowak
🔥 OUT: Memory Bank UI is now on Vertex AI Agent Engine!
Memory Bank provides persistent memory for agents.
Now, Vertex AI has introduced a full UI to manage it! 🥳
You can visually View 👁️, Edit ✏️, and Delete 🗑️ all agent memories in a unique place.
Check it out in the Agent Engine console! 👇
Memory Bank provides persistent memory for agents.
Now, Vertex AI has introduced a full UI to manage it! 🥳
You can visually View 👁️, Edit ✏️, and Delete 🗑️ all agent memories in a unique place.
Check it out in the Agent Engine console! 👇
September 12, 2025 at 4:44 PM
🔥 OUT: Memory Bank UI is now on Vertex AI Agent Engine!
Memory Bank provides persistent memory for agents.
Now, Vertex AI has introduced a full UI to manage it! 🥳
You can visually View 👁️, Edit ✏️, and Delete 🗑️ all agent memories in a unique place.
Check it out in the Agent Engine console! 👇
Memory Bank provides persistent memory for agents.
Now, Vertex AI has introduced a full UI to manage it! 🥳
You can visually View 👁️, Edit ✏️, and Delete 🗑️ all agent memories in a unique place.
Check it out in the Agent Engine console! 👇
Reposted by Alexander Nowak
A2A Extensions: Customize your multi-agent systems.
The A2A protocol is great, but what if you need something more tailored? A2A Extensions allow you to add custom, domain-specific functionalities to your agents.
Check out the blog to learn more:
developers.googleblog.com/en/a2a-exte...
The A2A protocol is great, but what if you need something more tailored? A2A Extensions allow you to add custom, domain-specific functionalities to your agents.
Check out the blog to learn more:
developers.googleblog.com/en/a2a-exte...
September 10, 2025 at 7:15 AM
A2A Extensions: Customize your multi-agent systems.
The A2A protocol is great, but what if you need something more tailored? A2A Extensions allow you to add custom, domain-specific functionalities to your agents.
Check out the blog to learn more:
developers.googleblog.com/en/a2a-exte...
The A2A protocol is great, but what if you need something more tailored? A2A Extensions allow you to add custom, domain-specific functionalities to your agents.
Check out the blog to learn more:
developers.googleblog.com/en/a2a-exte...
Reposted by Alexander Nowak
We've just released an amazing Embedding model:
EmbeddingGemma, the new best-in-class open embedding model! 🚀
🏆 Top multilingual model on MTEB (<500M)
💾 Runs on <200MB RAM
⚙️ Customizable output for on-device use
🧩 Integrated with your favorite tools
developers.googleblog.com/en/introduci...
EmbeddingGemma, the new best-in-class open embedding model! 🚀
🏆 Top multilingual model on MTEB (<500M)
💾 Runs on <200MB RAM
⚙️ Customizable output for on-device use
🧩 Integrated with your favorite tools
developers.googleblog.com/en/introduci...
Introducing EmbeddingGemma: The Best-in-Class Open Model for On-Device Embeddings- Google Developers Blog
Discover EmbeddingGemma, Google's new on-device embedding model designed for efficient on-device AI, enabling features like RAG and semantic search.
developers.googleblog.com
September 4, 2025 at 5:17 PM
We've just released an amazing Embedding model:
EmbeddingGemma, the new best-in-class open embedding model! 🚀
🏆 Top multilingual model on MTEB (<500M)
💾 Runs on <200MB RAM
⚙️ Customizable output for on-device use
🧩 Integrated with your favorite tools
developers.googleblog.com/en/introduci...
EmbeddingGemma, the new best-in-class open embedding model! 🚀
🏆 Top multilingual model on MTEB (<500M)
💾 Runs on <200MB RAM
⚙️ Customizable output for on-device use
🧩 Integrated with your favorite tools
developers.googleblog.com/en/introduci...