💼 Full Stack Engineer (Enterprise) @roboflow.
Build computer vision features with Jetson hardware.
🔧 JavaScript, Node.js, Python, Docker, Linux
📍 NY, SF or Remote - US
💰 $165K-$195K
https://t.co/YSYmSmf7Iv
#Python #ComputerVision #RemoteJobs #NYCJobs #SFJobs #TechJobs #Node
Did object detection just become a prompting problem?
The headline is Astra’s benchmark lead. The more consequential result is how it achieved it. On Roboflow Vision Evals, GPT-6 Astra reached 82.1% mAP@50 at low reasoning effort—5.4 points above Qwen3.8 Max and 13.7 above
GPT-6 Astra is the best vision model
Ok this is pretty insane. Astra side-loaded a custom app for my DJI RC2 controller that takes a live video stream from my Mini 5 Pro drone, streams it to a @roboflow Workflow running RF-DETR via our Serverless Video API, & overlays the predictions back on the controller’s screen. https://t.co/Y3XHp05JYe
Grok 4.7 is now available in @roboflow Workflows and Playgrounds
wanna try it on your own images? run it side-by-side with other vision models
link: https://t.co/YYwGMfNncq https://t.co/muQiWKzENc
5/6
The renderer passed full repo verification with 2,836 Vitest tests.
The builder fix passed 411 HTTP tests plus repeated browser checks.
Both PRs were merged, and a Roboflow maintainer confirmed the builder fix for release.
Look what we found! 👀 @IAStartupWeek is running a BOGO sale! If you want to embed yourself in the Iowa startup scene next week, get your tickets!
Justin will be interviewing the founder of @roboflow live!
https://t.co/sPBwrzQHmU
🚀 Trending on GitHub
supervision by roboflow
⭐ 50,987 stars
🧠 A reusable Python toolkit for computer vision: detection, tracking, and annotation blocks that plug into modern vision and multimodal AI pipelines.
https://t.co/1gaik2TbQT
#GenAI #OpenSource #AI
I’d recommend everyone working on multimodality to read this Roboflow post on GPT-6 Astra:
https://t.co/asbakrguL2
After working on multimodality for half a year, I increasingly feel that some of the most important problems are still the old ones:
OCR. Grounding. Segmentation.
The stack:
~30 frames semi-supervised on @roboflow, then RF-DETR segmentation every frame
@AIatMeta's SAM 2, with the modified memory bank
@opencvlibrary geometry for the camera motion, no net, no training
a small OCR pass on a few crops for long gaps, nothing jersey-specific
Supervision is a model-agnostic Python computer-vision toolkit with a unified detections representation and connectors for Ultralytics, Transformers, MMDetection, and Roboflow Inference. It handles datasets, model outputs, image and video annotation, tracking, and real-time zone
Rubberneck - ask your city's street cameras.
Coming soon to a Seattle near you :-)
Made with @typesafeai Jev, @cactuscompute Needle, and @roboflow RF-DETR. https://t.co/YMH7TfYxWx
roboflow/supervision handles the messy real-world data part that most vision agents still choke on. If you're shipping anything that sees camera feeds, start here instead of writing another preprocessing script.
Astra now powers Roboflow Auto Annotate
we asked for Celtics and Knicks players using team names alone
the same prompt worked across different arenas, uniforms, lighting and players
try your images for free: https://t.co/KxJxSsVmBu https://t.co/Rt2hhJqaW7
🔍 Roboflow's segmentation recipe: use Astra for detection, SAM3 for masks. Astra understands complex text prompts SAM3 misses, but draws pricey JSON polygons (~$0.50 per hi-res image); boxes from Astra plus SAM3 edges cut that ~80%, per Roboflow.
Their measured pure-VLM run: https://t.co/vq6x95KLWh
GPT-6 Astra: The Best Vision Model Tested on @roboflow
@skalskip92 covered object detection, text and box prompting, segmentation, auto-labeling, and re-identification. Also touch on vision-language-action models (VLAs) and how Astra could be used in systems that connect vision https://t.co/g9pbqUnAKv
📛 supervision
🧠 Supervision is a computer vision toolkit for model integration, customizable visual annotations, and dataset management.
🛠️ @roboflow
💻 Python
⭐ 50780
🍴 4827
🔎 roboflow/supervision on GitHub
Plant cameras already run. The missing piece is reusable CV plumbing.
50,751★ · +327 today · #14 on GitHub Trending.
roboflow/supervision: model-agnostic detect → track → annotate → zone count — plug Ultralytics / Transformers / Inference and ship the shop bay, not another https://t.co/9Et0rB3qdv
Score Studio is now live, giving independent engineers and small teams an alternative to Roboflow for building computer vision systems.
They can generate and label data, use the subnet’s public-track models or train their own, and compare results in one workspace.
A visual https://t.co/WYnBnlwuCw
fresh from @roboflow: Breathing detection web application — made with GPT-6 Astra
find more at https://t.co/J4GUK22rF3 https://t.co/YP4esJ8c72
📛 supervision
🧠 Supervision is a Python toolkit providing dataset management, model connectors, and customizable annotators for computer vision.
🛠️ @roboflow
💻 Python
⭐ 50568
🍴 4812
🔎 roboflow/supervision on GitHub
Some more examples. I think this one is really good using @roboflow
https://t.co/B5JERIS0mq https://t.co/y1LTi7nCj1
Some more examples. I think this one is really good using @roboflow
https://t.co/B5JERIS0mq https://t.co/y1LTi7nCj1
a free computer vision library crossed 50,000 github stars this week, and the maintainer's post said thank you instead of announcing a pricing page
supervision, by Piotr Skalski at Roboflow. MIT licence, 5,131 commits, 1,732 pull requests, 188 contributors, started in november https://t.co/8BmBMwzAq3
Inspired by R2-D2 🤖, I’m building Reachy Mini into a robot whose perception, memory, and movement work together as one system.
It can map in RGB-D, detect and remember objects, search for people, plan paths, and follow them while adapting in real time.
Works on top of https://t.co/hV8RDDwGPb
Roboflow users are already on a >$1M/yr run-rate for Astra auto-label consumption one week in. It seems the people like it and are getting great results! https://t.co/7BPqcpwzek