Cos (@CosWiSe) header
Cos (@CosWiSe) profile picture

Senior AI Product Designer // Solving problems using AI-Driven Solutions // Product Experience // Agentic Architect // Proud nerd

London, England
Joined September 2010
Photo by Cos (@CosWiSe)Photo by Cos (@CosWiSe)Photo by Cos (@CosWiSe)Photo by Cos (@CosWiSe)Photo by Cos (@CosWiSe)Photo by Cos (@CosWiSe)Photo by Cos (@CosWiSe)Photo by Cos (@CosWiSe)Photo by Cos (@CosWiSe)Photo by Cos (@CosWiSe)
DESIGN.md Websites. 1. typeui.sh 2. designmd.me 3. designmd.supply 4. styles.refero.design 5. getdesign.md 6. collectui.com 7. designmd-store.com What did I miss?
25
204
3
1,803
115,679
oh myโ€ฆ this shouldnโ€™t be possible Gemini 3 can generate 3D interactive scenes with three.jsโ€ฆ and you can literally move particles with your hands no coding skills needed at all, it's all free tutorial + prompts in the comments
267
1,129
134
11,392
1,267,268
If you listen very very closely, you can hear engineers screaming.
22
22
4
383
26,256
good morning everyone ๐Ÿ‘๏ธ
9
40
275
6,478
326
316
53
4,367
193,144
๐Ÿ’š So proud to have been nominated, we'd love your support!
This Post is from an account that no longer exists. Learn more
1
3
25
5,705
Designers when you say you'll pay them with equity and exposure
8
20
4
206
19,587
7,455
37,918
3,452
251,751
38,964,247
wake up babe Figma to React is real!
52
314
69
3,442
504,025
Life Hack: if there is a โ€œcustomโ€ tip option you can sometimes enter a negative number and make your meal cheaper
607
2,072
309
76,556
9,795,274
Designers know
19
173
23
1,574
105,779
HOLY SHIT -- @OpenAI increased the rate limits! Limits for GPT3.5 have gone up ~5x ๐Ÿคฏ๐Ÿคฏ๐Ÿคฏ
20
24
3
335
79,453
Hiring๐Ÿ”ฅ Junior Frontend Developer๐Ÿ› ๏ธ Location: Remote/Hybrid ๐ŸŒ Compensation: โ‚ฌ65,000๐Ÿ’ฐ Stack: HTML, CSS, JavaScript Apply: showwcase.com/job/9007
18
124
6
709
126,228
Job market for creatives, 2024: - Model training art director - Fine-tuning specialist - Video generation engineer - Associate dataset curator - AI director - AI Camera operator
27
32
12
266
71,339
Interested in NLP? Gone are the days when NLP engineers trained models from scratch. Fine-tuning is the new training. Here are 2 open-source Python libraries to help you ๐Ÿโ†“
3
29
1
169
39,073
Do Large Language Models have any practical value? I'm meeting people daily, and around 90% of them tell me they can't scale Large Language Models past a cool-looking demo. The models sound impressive but fail spectacularly whenever you try to build anything serious with them. I think there's some truth here, but the problem goes a bit deeper: Most of these companies don't know how to use these models. You need a lot to build a working application: data connectors, embeddings, vector databases, agents, and models. One of the most common use cases is using a model to ask questions about your data. Imagine you are a law firm and want to research thousands of settled case law documents. How can you use these models to solve this problem? If you are stuck here, I want to give you a good idea: I built a simple example using LlamaIndex and @monsterapis. LlamaIndex is an open-source framework for building language-based applications, and @monsterapis will give you access to every open-source Large Language Model in the market right now. With LlamaIndex, you can orchestrate an end-to-end application. They offer a hub of data loaders and agent tools that you can use to build anything you can imagine. If you have data and want to use an LLM with it, LlamaIndex will help and hide most of the complexity from you. My example code will help you build an application to ask questions about your data. In 15 lines of code, I index the contents of a folder and use Llama 2 to ask questions about it. I've been working with @monsterapis for a while, so I asked them to partner with me to use their official integration with LlamaIndex. This is the perfect combination! Here is the Colab notebook that you can run: colab.research.google.com/drโ€ฆ Have you used LlamaIndex before? What are your thoughts?
48
162
6
996
345,412
Future CSS Tip! โšก๏ธ You can combine CSS :has() & :user-valid/:user-invalid to power <form> micro interactions ๐Ÿ˜ No JavaScript. Watch for the little animations โœจ label { --color: var(--default); color: var(--color); border-color: var(--color); } .group:has(:user-valid) { --color: var(--valid); } .group:not(:focus-within):has(:user-invalid) { --color: var(--invalid); } These new pseudo-elements only fire once an input has been interacted with. This means you don't need to do things like the "transparent :placeholder-shown" hack from before ๐Ÿ™Œ Then you can lean into the power of scoped custom properties and change the theme based on the status of a form group ๐Ÿ˜Ž It's a nice little touch. But remember, don't rely on color to communicate things! @CodePen link below! ๐Ÿ‘‡
Blink: Intent to Ship: :user-valid and :user-invalid CSS pseudo-classes groups.google.com/a/chromiumโ€ฆ
16
308
7
1,921
284,071
What does an ๐—˜๐—ณ๐—ณ๐—ฒ๐—ฐ๐˜๐—ถ๐˜ƒ๐—ฒ ๐— ๐—ฎ๐—ฐ๐—ต๐—ถ๐—ป๐—ฒ ๐—Ÿ๐—ฒ๐—ฎ๐—ฟ๐—ป๐—ถ๐—ป๐—ด ๐—˜๐˜…๐—ฝ๐—ฒ๐—ฟ๐—ถ๐—บ๐—ฒ๐—ป๐˜๐—ฎ๐˜๐—ถ๐—ผ๐—ป ๐—˜๐—ป๐˜ƒ๐—ถ๐—ฟ๐—ผ๐—ป๐—บ๐—ฒ๐—ป๐˜ look like? MLOps practices are there to improve Machine Learning Product development velocity, the biggest bottlenecks happen when Experimentation Environments and other infrastructure elements are integrated poorly. Letโ€™s look into the properties that an effective Experimentation Environment should have. As a MLOps engineer you should strive to provide these to your users and as a Data Scientist, you should know what you should be demanding for. ๐Ÿญ: Access to the raw data. While handling raw data is the responsibility of Data Engineering function, Data Scientists need the ability to explore and analyze available raw data and decide which of it needs to be moved upstream the Data Value Chain (2.1). ๐Ÿฎ: Access to the curated data. Curated data might be available in the Data Warehouse but not exposed via a Feature Store. Such Data should not be exposed for model training in production environments. Data Scientists need the ability to explore curated data and see what needs to be pushed downstream (3.1). ๐Ÿฏ: Data used for training of Machine Learning models should be sourced from a Feature Store if the ML Training pipeline is ready to be moved to the production stage. ๐Ÿฐ: Data Scientists should be able to easily spin up different types of compute clusters - might it be Spark, Dask or any other technology - to allow effective Raw and Curated Data exploration. ๐Ÿฑ: Data Scientists should be able to spin up a production like remote Machine Learning Training pipeline in development environment ad-hoc from the Notebook, this increases speed of iteration significantly. ๐Ÿฒ: There should be an automated setup in place that would perform the testing and promotion to a higher env when a specific set of Pull Requests are created. E.g. a PR from feature/* to release/* branch could trigger a CI/CD process to test and deploy the ML Pipeline to a pre-prod environment. ๐Ÿณ: Notebooks and any additional boilerplate code for CI/CD should be part of your Git integration. Make it crystal clear where a certain type of code should live - a popular way to do this is providing repository templates with clear documentation. ๐Ÿด: Experiment/Model Tracking System should be exposed to both local and remote pipelines. ๐Ÿ—: Notebooks have to be running in the same environment that your production code will run in. Incompatible dependencies should not cause problems when porting applications to production. It can be achieved by running Notebooks in containers. Did I miss something? ๐Ÿ‘‡ -------- Follow me to upskill in #MLOps, #MachineLearning, #DataEngineering, #DataScience and overall #Data space. Also hit ๐Ÿ””to stay notified about new content. ๐——๐—ผ๐—ปโ€™๐˜ ๐—ณ๐—ผ๐—ฟ๐—ด๐—ฒ๐˜ ๐˜๐—ผ ๐—น๐—ถ๐—ธ๐—ฒ ๐Ÿ’™, ๐˜€๐—ต๐—ฎ๐—ฟ๐—ฒ ๐—ฎ๐—ป๐—ฑ ๐—ฐ๐—ผ๐—บ๐—บ๐—ฒ๐—ป๐˜! Join a growing community of Data Professionals by subscribing to my ๐—ก๐—ฒ๐˜„๐˜€๐—น๐—ฒ๐˜๐˜๐—ฒ๐—ฟ: newsletter.swirlai.com
2
62
1
219
19,707
For the past few months, I've become obsessed with the problem of how do you build an AI-powered business AKA - "how do we actually use this stuff" But it honestly started because at the beginning of the year, I did hundreds of AI Advising calls with businesses and while I was able to answer questions and help point them in the right direction... The meaty question they really needed answered was - "How do I use AI to power my business" Pre-ChatGPT, that wasn't really a question on people's radar. Much less one that had good answers. (Remember, using AI used to be more of a FAANG thing) Underneath the question are all sorts of sub-questions... - What should people learn - Who should businesses hire - Where should businesses start - Why do certain things - Why not do other things - What ethical guardrails should exist + more I think we're still at Day 1 in answering these questions. But I'm also proud of the progress. (Posted screenshot of part of this morning's discussion on it below) It's early for AI Operations. But definition breeds clearer investment, focus and is an important step for getting real world business value out of AI. Btw, our @theaiexchange slack is where it's happening. 1200+ members discussing, shaping, sharing stories and use cases for using AI in work & operating businesses ๐Ÿ’œ
42
41
4
461
253,209