Anton Dorozhko

Senior SWE, ML, and Solutions at
#MLOps #DataCentricAI #ComputerVision

[Anton Dorozhko @dorozhkoa]

Checkout my new tutorial about Streaming Object detection metadata from Deepstream to Kafka in 2 lines of code https://t.co/pXPoFedgCx

#computervision #deepstream #nvidia

[Anton Dorozhko @dorozhkoa]

and many billions of inferences on Edge AI devices on the Nvidia Jetson platform and server-side GPUs. In Smart Cities, Smart Spaces, and Industrial QA

Let’s Build Intelligent Solutions Together!

Book a Value Discovery call here https://t.co/E5wH7WqnY8

[Anton Dorozhko @dorozhkoa]

This year I broadened my offers by creating Value Stream IE.

I’m looking for Computer Vision Engineer roles and/or new projects for ValueStream.

My solutions served to process:
100s of concurrent real-time video streams
30+ different computer vision models
....

[Anton Dorozhko @dorozhkoa]

Do you see this "blue" school bus with the number "27B" and the word "SCHOOL"?

Gladly it is NOT a VLM Hallucitation!
It's just a bug in NVIDIA video summarization sample on Jetson that is very common in video analytics.

walkthough video: https://t.co/T1B6QnDoy0

[Anton Dorozhko @dorozhkoa]

Zero-shot detection with NanoOWL on Jetson devices. In this video, we cover a video processing pipeline to deploy open-vocabulary object detection that can handle even objects for which you don't have training data.
#zeroshot #nanoowl #nvidia #jetson
https://t.co/gm4CWb1uow

[Anton Dorozhko @dorozhkoa]

@Nvidia 's NVstreamer saved me quite some time by easily creating RTSP streams from your videos for impressive live demos or testing your video processing pipelines in deepstream or other frameworks.

Learn how to setup it on jetson and servers
https://t.co/XbQ3SqnuLw

[Deedy @deedydas]

HUGE Immigration News for International Entrepreneurs!!

If you own 10%+ of a US startup entity founded <5yrs ago with $264k+ of funding, you+spouses of up to 3 co-founders can come work in the US for 2.5yrs with renewal to 5yrs.

Startups globally can now come build in SF!

1/5

[ludwig @ludwigABAP]

daily reminder:

all these "big scary maths symbols are actually simpler to understand in code" people need to read "A Programmer's introduction to Mathematics" by Jeremy Kun

The whole book is about that

[Thomas Kipf @tkipf]

Absolutely wild

[Andrej Karpathy @karpathy]

LLM OS. Bear with me I'm still cooking.

Specs:
- LLM: OpenAI GPT-4 Turbo 256 core (batch size) processor @ 20Hz (tok/s)
- RAM: 128Ktok
- Filesystem: Ada002

[Weights & Biases @wandb]

Exploring the future of AI in vision systems?🔎

Join us for a virtual session with @soumikRakshit96 & @dorozhkoa from @Smartcow_ai.

Get insights on Ultralytics, object detection techniques, and engage in a live Q&A session.💬

Save your seat today 🔗 https://t.co/Mwx2nYCnTB

[Weights & Biases @wandb]

Dive into AI & computer vision with our virtual session featuring @soumikRakshit96 & @dorozhkoa from @Smartcow_ai.

Learn about Ultralytics, object detection, & more. Don't miss the interactive Q&A!💭

Reserve your spot now🔗 https://t.co/Mwx2nYCnTB

[GeekyRakshit (e/smol_model) @soumikRakshit96]

📸 I will be joining @dorozhkoa in speaking about developing computer vision systems using @ultralytics and @wandb on 8 Nov, 5:30 PM CET.

👉 Reserve your spot at https://t.co/ZMNXs0H1Vj

[Weights & Biases @wandb]

Dive into AI & computer vision with our virtual session featuring @soumikRakshit96 & @dorozhkoa from @Smartcow_ai.🔬

Learn about Ultralytics, object detection, & more. Don't miss the interactive Q&A!💭

Reserve your spot now🔗 https://t.co/Mwx2nYCnTB

[Andrej Karpathy @karpathy]

With many 🧩 dropping recently, a more complete picture is emerging of LLMs not as a chatbot, but the kernel process of a new Operating System. E.g. today it orchestrates:

- Input & Output across modalities (text, audio, vision)
- Code interpreter, ability to write & run programs
- Browser / internet access
- Embeddings database for files and internal memory storage & retrieval

A lot of computing concepts carry over. Currently we have single-threaded execution running at ~10Hz (tok/s) and enjoy looking at the assembly-level execution traces stream by. Concepts from computer security carry over, with attacks, defenses and emerging vulnerabilities.

I also like the nearest neighbor analogy of "Operating System" because the industry is starting to shape up similar:
Windows, OS X, and Linux <-> GPT, PaLM, Claude, and Llama/Mistral(?:)).
An OS comes with default apps but has an app store.
Most apps can be adapted to multiple platforms.

TLDR looking at LLMs as chatbots is the same as looking at early computers as calculators. We're seeing an emergence of a whole new computing paradigm, and it is very early.

[Anton Dorozhko @dorozhkoa]

Building software we have infinite possibilities of implementing the same functionality. To efficiently navigate all those choices and decisions on a way to deliver value for our clients we need a solid first-principle understanding of modern practices.
https://t.co/4otfwmjHPQ

[Anton Dorozhko @dorozhkoa]

🏙️ How computer vision can help in improving 🏰 safety and 📹 security of our cities, increase a traffic flow 🌊 by detecting road conditions and different vehicle/pedestrian conflicts 🏎 🤺

📺 Watch Now  https://t.co/jWpqlv6uzJ  #SmartCities #ComputerVision

[Anton Dorozhko @dorozhkoa]

🏙️ How computer vision can help in maintaining transport 🚗 and power infrastructure ⚡ , improve the safety 🚔 of streets, and make a difference between vandalism 👿 and street art 🎨.

📺 Watch Now https://t.co/7O1sVDdgl8  #SmartCities #ComputerVision

[Anton Dorozhko @dorozhkoa]

🏙️ How #computervision can be applied in #SmartCity both 🚆 above and 🚇 under the ground, in the air 🛸 , and inside the buildings 🏢 ?

🔍 Watch the next 12 applications of an exciting series: "50 Computer Vision Applications for Smart City"! 🌆 https://t.co/jrTr1YYhTj

[Julia Gusak @juliagusak502]

Excited to announce our Workshop on Advancing Neural Network Training, WANT @NeurIPSConf!🚀 Save GPU hours, keep accuracy! Join HPC &amp; AI experts on Dec 16, submit papers by Sep 29. #AI #WANT #HPC #NeurIPS2023 #EfficientTraining Details: https://t.co/H5PIxu6xGw