Fosstodon

Domain-Driven Design EuropeJoin us for hands-on Machine Learning deployment training! You'll analyse errors, tweak models, and push to production using real-world engineering patterns—way beyond the "drop your model on S3 and call it a day" approach. Gain practical experience with sophisticated ML engineering techniques that you can immediately apply on the job. <a href="https://m.aardling.social/tags/MLEngineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#MLEngineering</a> <a href="https://m.aardling.social/tags/AITraining" class="mention hashtag" rel="nofollow noopener" target="_blank">#AITraining</a> 👉 <a href="https://ddd.academy/put-an-ml-model-in-production/" rel="nofollow noopener" translate="no" target="_blank">https://ddd.academy/put-an-ml-model-in-production/</a>

blaze.email📊 Dive deep with MIT's flow matching course or scale neural networks across thousands of GPUs with Jeremy Jordan's guide. The math behind efficient ML! <a href="https://mastodon.social/tags/DataScience" class="mention hashtag" rel="nofollow noopener" target="_blank">#DataScience</a> <a href="https://mastodon.social/tags/MLEngineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#MLEngineering</a> <a href="https://blaze.email/Machine-Learning-Engineer" rel="nofollow noopener" translate="no" target="_blank">https://blaze.email/Machine-Learning-Engineer</a>

MLE PathEarly in your ML career, every decision feels irreversible. But the best engineers don’t aim for perfection—they build with reversibility in mind.Understanding the difference between one-way and two-way doors will help you iterate faster and build better.<a href="https://mastodon.social/tags/MachineLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#MachineLearning</a> <a href="https://mastodon.social/tags/MLEngineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#MLEngineering</a> <a href="https://mastodon.social/tags/TechCareers" class="mention hashtag" rel="nofollow noopener" target="_blank">#TechCareers</a>

Judith van StegerenA few tips for optimizing Pytorch model training time from a Yandex ML engineer.<a href="https://alexdremov.me/simple-ways-to-speedup-your-pytorch-model-training/" target="_blank" rel="nofollow noopener" translate="no">https://alexdremov.me/simple-ways-to-speedup-your-pytorch-model-training/</a><a href="https://fosstodon.org/tags/ml" class="mention hashtag" rel="tag">#ml</a> <a href="https://fosstodon.org/tags/mlengineering" class="mention hashtag" rel="tag">#mlengineering</a> <a href="https://fosstodon.org/tags/modeltraining" class="mention hashtag" rel="tag">#modeltraining</a> <a href="https://fosstodon.org/tags/pytorch" class="mention hashtag" rel="tag">#pytorch</a> <a href="https://fosstodon.org/tags/modeloptimization" class="mention hashtag" rel="tag">#modeloptimization</a>

Vic5 in-demand ☁️ <a href="https://techhub.social/tags/cloudjobs" class="mention hashtag" rel="nofollow noopener" target="_blank">#cloudjobs</a> for 2024: 1➧ Cloud Security Engineer: $152,000. Responsibilities include securing digital estates and protecting networks. 2➧ Cloud Solutions Architect: $150,000. Create and maintain cloud solutions. 3➧ Cloud AI/ML Engineer: $162,000. Build AI and ML systems for data-driven decision-making. 4➧ DevOps Engineer: $123,000. Focusing on cloud infrastructure and IT operations automation. 5➧ Cloud Support Engineer: $130,000. Acting as liaisons for cloud products and customer support.<a href="https://www.sdxcentral.com/articles/analysis/5-high-paying-and-in-demand-cloud-jobs-for-2024/2024/01/" rel="nofollow noopener" translate="no" target="_blank">https://www.sdxcentral.com/articles/analysis/5-high-paying-and-in-demand-cloud-jobs-for-2024/2024/01/</a> <a href="https://techhub.social/tags/CloudComputing" class="mention hashtag" rel="nofollow noopener" target="_blank">#CloudComputing</a> <a href="https://techhub.social/tags/CloudJobs" class="mention hashtag" rel="nofollow noopener" target="_blank">#CloudJobs</a> <a href="https://techhub.social/tags/TechCareers" class="mention hashtag" rel="nofollow noopener" target="_blank">#TechCareers</a> <a href="https://techhub.social/tags/CloudSecurity" class="mention hashtag" rel="nofollow noopener" target="_blank">#CloudSecurity</a> <a href="https://techhub.social/tags/SolutionsArchitecture" class="mention hashtag" rel="nofollow noopener" target="_blank">#SolutionsArchitecture</a> <a href="https://techhub.social/tags/AI" class="mention hashtag" rel="nofollow noopener" target="_blank">#AI</a> <a href="https://techhub.social/tags/AIEngineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#AIEngineering</a> <a href="https://techhub.social/tags/ML" class="mention hashtag" rel="nofollow noopener" target="_blank">#ML</a> <a href="https://techhub.social/tags/MLEngineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#MLEngineering</a> <a href="https://techhub.social/tags/DevOps" class="mention hashtag" rel="nofollow noopener" target="_blank">#DevOps</a> <a href="https://techhub.social/tags/CloudSupport" class="mention hashtag" rel="nofollow noopener" target="_blank">#CloudSupport</a> <a href="https://techhub.social/tags/TechTrends" class="mention hashtag" rel="nofollow noopener" target="_blank">#TechTrends</a> <a href="https://techhub.social/tags/Innovation" class="mention hashtag" rel="nofollow noopener" target="_blank">#Innovation</a> <a href="https://techhub.social/tags/DataSecurity" class="mention hashtag" rel="nofollow noopener" target="_blank">#DataSecurity</a> <a href="https://techhub.social/tags/MachineLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#MachineLearning</a> <a href="https://techhub.social/tags/ArtificialIntelligence" class="mention hashtag" rel="nofollow noopener" target="_blank">#ArtificialIntelligence</a><a href="https://techhub.social/tags/FrontierAirlines" class="mention hashtag" rel="nofollow noopener" target="_blank">#FrontierAirlines</a> <a href="https://techhub.social/tags/Deloitte" class="mention hashtag" rel="nofollow noopener" target="_blank">#Deloitte</a> <a href="https://techhub.social/tags/Intel" class="mention hashtag" rel="nofollow noopener" target="_blank">#Intel</a> <a href="https://techhub.social/tags/Hinge" class="mention hashtag" rel="nofollow noopener" target="_blank">#Hinge</a> <a href="https://techhub.social/tags/Microsoft" class="mention hashtag" rel="nofollow noopener" target="_blank">#Microsoft</a> <a href="https://techhub.social/tags/PaloAltoNetworks" class="mention hashtag" rel="nofollow noopener" target="_blank">#PaloAltoNetworks</a> <a href="https://techhub.social/tags/Infosys" class="mention hashtag" rel="nofollow noopener" target="_blank">#Infosys</a> <a href="https://techhub.social/tags/KeysightTechnologies" class="mention hashtag" rel="nofollow noopener" target="_blank">#KeysightTechnologies</a> <a href="https://techhub.social/tags/Atlassian" class="mention hashtag" rel="nofollow noopener" target="_blank">#Atlassian</a> <a href="https://techhub.social/tags/Netflix" class="mention hashtag" rel="nofollow noopener" target="_blank">#Netflix</a> <a href="https://techhub.social/tags/DellTechnologies" class="mention hashtag" rel="nofollow noopener" target="_blank">#DellTechnologies</a> <a href="https://techhub.social/tags/Salesforce" class="mention hashtag" rel="nofollow noopener" target="_blank">#Salesforce</a> <a href="https://techhub.social/tags/FICO" class="mention hashtag" rel="nofollow noopener" target="_blank">#FICO</a> <a href="https://techhub.social/tags/Gartner" class="mention hashtag" rel="nofollow noopener" target="_blank">#Gartner</a> <a href="https://techhub.social/tags/TheHomeDepot" class="mention hashtag" rel="nofollow noopener" target="_blank">#TheHomeDepot</a> <a href="https://techhub.social/tags/AmazonWebServices" class="mention hashtag" rel="nofollow noopener" target="_blank">#AmazonWebServices</a> <a href="https://techhub.social/tags/DigitalOcean" class="mention hashtag" rel="nofollow noopener" target="_blank">#DigitalOcean</a> <a href="https://techhub.social/tags/Snowflake" class="mention hashtag" rel="nofollow noopener" target="_blank">#Snowflake</a> <a href="https://techhub.social/tags/Couchbase" class="mention hashtag" rel="nofollow noopener" target="_blank">#Couchbase</a>

Chris OffnerHere's a more clearly visible demonstration of the problem I described previously: <a href="https://sigmoid.social/@chrisoffner3d/111591367887994819" rel="nofollow noopener" translate="no" target="_blank">https://sigmoid.social/@chrisoffner3d/111591367887994819</a>On the left we see the progression of cross-attention maps extracted via the CPU, on the right we see the same cross-attention maps extracted via the GPU.This is using the <a href="https://sigmoid.social/tags/Keras" class="mention hashtag" rel="nofollow noopener" target="_blank">#Keras</a> implementation of <a href="https://sigmoid.social/tags/StableDiffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#StableDiffusion</a> on an M3 Max.<a href="https://sigmoid.social/tags/TensorFlow" class="mention hashtag" rel="nofollow noopener" target="_blank">#TensorFlow</a> <a href="https://sigmoid.social/tags/StableDiffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#StableDiffusion</a> <a href="https://sigmoid.social/tags/Diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#Diffusion</a> <a href="https://sigmoid.social/tags/Python" class="mention hashtag" rel="nofollow noopener" target="_blank">#Python</a> <a href="https://sigmoid.social/tags/MLEngineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#MLEngineering</a> <a href="https://sigmoid.social/tags/MachineLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#MachineLearning</a> <a href="https://sigmoid.social/tags/DeepLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#DeepLearning</a> <a href="https://sigmoid.social/tags/GPU" class="mention hashtag" rel="nofollow noopener" target="_blank">#GPU</a> <a href="https://sigmoid.social/tags/M3Max" class="mention hashtag" rel="nofollow noopener" target="_blank">#M3Max</a>

Chris OffnerFor example, check the second row, fifth column and how it changes between t = 600 and t = 700.Is this some bug specific to Apple GPUs or does this also happen with CUDA?For t = 0, the CPU and GPU images look identical. For higher t, the GPU run produces *very* different results even when re-running with the exact same model inputs, i.e. also for the same time step t.Any idea why that is?<a href="https://sigmoid.social/tags/MLEngineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#MLEngineering</a> <a href="https://sigmoid.social/tags/GPU" class="mention hashtag" rel="nofollow noopener" target="_blank">#GPU</a> <a href="https://sigmoid.social/tags/DeepLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#DeepLearning</a> <a href="https://sigmoid.social/tags/Diffusion" class="mention hashtag" rel="nofollow noopener" target="_blank">#Diffusion</a> <a href="https://sigmoid.social/tags/CUDA" class="mention hashtag" rel="nofollow noopener" target="_blank">#CUDA</a> <a href="https://sigmoid.social/tags/AppleSilicon" class="mention hashtag" rel="nofollow noopener" target="_blank">#AppleSilicon</a> <a href="https://sigmoid.social/tags/TensorFlow" class="mention hashtag" rel="nofollow noopener" target="_blank">#TensorFlow</a> <a href="https://sigmoid.social/tags/Keras" class="mention hashtag" rel="nofollow noopener" target="_blank">#Keras</a>

Chris OffnerI'm running into some unexpected and significant non-determinism when running a <a href="https://sigmoid.social/tags/Keras" class="mention hashtag" rel="nofollow noopener" target="_blank">#Keras</a> diffusion model on my Apple GPU.On the left we see the progression of cross-attention maps for time steps from t = 0 to t = 900 when running the model via the CPU.We see that each cross-attention map undergoes some "refinement" progression as we go from t = 0 to t= 900.On the right we see the same but on the GPU.It's a much more erratic and discontinuous progression.<a href="https://sigmoid.social/tags/MLEngineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#MLEngineering</a> <a href="https://sigmoid.social/tags/DeepLearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#DeepLearning</a> <a href="https://sigmoid.social/tags/GPU" class="mention hashtag" rel="nofollow noopener" target="_blank">#GPU</a>

Mikiko🔖 The Top 5 Papers About <a href="https://data-folks.masto.host/tags/mlops" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlops</a> You Should Know (Part 1)I've seen a ton of lists about the most important papers in <a href="https://data-folks.masto.host/tags/ml" class="mention hashtag" rel="nofollow noopener" target="_blank">#ml</a>, <a href="https://data-folks.masto.host/tags/datascience" class="mention hashtag" rel="nofollow noopener" target="_blank">#datascience</a>, <a href="https://data-folks.masto.host/tags/deeplearning" class="mention hashtag" rel="nofollow noopener" target="_blank">#deeplearning</a>, <a href="https://data-folks.masto.host/tags/mlengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlengineering</a>. But I've either seen not that many <a href="https://data-folks.masto.host/tags/mlops" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlops</a> reading lists or when I do run across them they tend to be focused a bit too deeply on specific ML systems or domains or algorithms. 👉🏻 If you only read 5 papers to understand why ML is hard (and how big the problem space of MLOps is) it should be these papers. [To Be Continued]

Tim KelloggDoes anyone here have experience with <a href="https://hachyderm.io/tags/Prefect" class="mention hashtag" rel="nofollow noopener" target="_blank">#Prefect</a>? What's the best way to automate blocks? can you do it via <a href="https://hachyderm.io/tags/terraform" class="mention hashtag" rel="nofollow noopener" target="_blank">#terraform</a>? <a href="https://hachyderm.io/tags/ml" class="mention hashtag" rel="nofollow noopener" target="_blank">#ml</a> <a href="https://hachyderm.io/tags/mlengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlengineering</a>

MikikoThe tools we have today are better than the ones we had before and this is especially true in the <a href="https://data-folks.masto.host/tags/mlops" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlops</a> world. We have more options than ever before (cc: MAD Turck Landscape) but confusion is just as high as it ever was.<a href="https://data-folks.masto.host/tags/mlops" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlops</a> <a href="https://data-folks.masto.host/tags/productionml" class="mention hashtag" rel="nofollow noopener" target="_blank">#productionml</a> <a href="https://data-folks.masto.host/tags/mlengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlengineering</a> <a href="https://data-folks.masto.host/tags/oss" class="mention hashtag" rel="nofollow noopener" target="_blank">#oss</a> <a href="https://data-folks.masto.host/tags/devtools" class="mention hashtag" rel="nofollow noopener" target="_blank">#devtools</a> <a href="https://data-folks.masto.host/tags/python" class="mention hashtag" rel="nofollow noopener" target="_blank">#python</a>

MikikoHaving <a href="https://data-folks.masto.host/tags/DataScientists" class="mention hashtag" rel="nofollow noopener" target="_blank">#DataScientists</a> Build Infrastructure & Developing Models At The Same Time Is A Terrible Anti-Pattern We’re Addicted To. Esp at comps that aren’t early stage -- correlated w/ a lack of technical DS leadership, poor infra design, and lack of organizational alignment. Really shows how the difference between success & failure isn’t technology choices but good project management & strategic leadership around platforms.<a href="https://data-folks.masto.host/tags/mlops" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlops</a> <a href="https://data-folks.masto.host/tags/mlengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlengineering</a> <a href="https://data-folks.masto.host/tags/mlplatforms" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlplatforms</a> <a href="https://data-folks.masto.host/tags/datascience" class="mention hashtag" rel="nofollow noopener" target="_blank">#datascience</a>

Mikiko👉🏻 Online Inference =/= StreamingWe're all aware of this right? That they're not the same thing?<a href="https://data-folks.masto.host/tags/mlops" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlops</a> <a href="https://data-folks.masto.host/tags/mlengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlengineering</a> <a href="https://data-folks.masto.host/tags/datascience" class="mention hashtag" rel="nofollow noopener" target="_blank">#datascience</a> <a href="https://data-folks.masto.host/tags/dataengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#dataengineering</a> <a href="https://data-folks.masto.host/tags/productionml" class="mention hashtag" rel="nofollow noopener" target="_blank">#productionml</a> <a href="https://data-folks.masto.host/tags/mlsystems" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlsystems</a> <a href="https://data-folks.masto.host/tags/systemdesign" class="mention hashtag" rel="nofollow noopener" target="_blank">#systemdesign</a>

Mikiko🤔 Rather than trying to get rid of the <a href="https://data-folks.masto.host/tags/datascientist" class="mention hashtag" rel="nofollow noopener" target="_blank">#datascientist</a> title, maybe we just treat it as an abstract class and continue on our merry ways? <a href="https://data-folks.masto.host/tags/datascience" class="mention hashtag" rel="nofollow noopener" target="_blank">#datascience</a> <a href="https://data-folks.masto.host/tags/dataengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#dataengineering</a> <a href="https://data-folks.masto.host/tags/mlops" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlops</a> <a href="https://data-folks.masto.host/tags/mlengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlengineering</a> <a href="https://data-folks.masto.host/tags/ai" class="mention hashtag" rel="nofollow noopener" target="_blank">#ai</a> <a href="https://data-folks.masto.host/tags/career" class="mention hashtag" rel="nofollow noopener" target="_blank">#career</a> <a href="https://data-folks.masto.host/tags/data" class="mention hashtag" rel="nofollow noopener" target="_blank">#data</a>

Mikiko🤔To bootcamp or not to bootcamp? Like all annoying senior devs, my answer is going to be: "It depends". I breakdown what consider when choosing the <a href="https://data-folks.masto.host/tags/bootcamp" class="mention hashtag" rel="nofollow noopener" target="_blank">#bootcamp</a> route for <a href="https://data-folks.masto.host/tags/datascience" class="mention hashtag" rel="nofollow noopener" target="_blank">#datascience</a> (but advice good for other bootcamps like <a href="https://data-folks.masto.host/tags/dataengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#dataengineering</a> <a href="https://data-folks.masto.host/tags/mlengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlengineering</a>, etc) #🐘 <a href="https://t.co/sTiiwWOB7D" rel="nofollow noopener" target="_blank">https://t.co/sTiiwWOB7D</a>

MikikoIf the answer is similar to: 1️⃣ ASAP 2️⃣ Minimal 3️⃣ Divorced 4️⃣ We can't 5️⃣ Less than 5Then your first step shouldn't be building an ML platform, it should be developing models or ML-drive product features using the simplest, tried & true patterns possible.<a href="https://data-folks.masto.host/tags/mlops" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlops</a> <a href="https://data-folks.masto.host/tags/mlplatform" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlplatform</a> <a href="https://data-folks.masto.host/tags/datascience" class="mention hashtag" rel="nofollow noopener" target="_blank">#datascience</a> <a href="https://data-folks.masto.host/tags/mlengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlengineering</a> <a href="https://data-folks.masto.host/tags/platformengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#platformengineering</a> <a href="https://data-folks.masto.host/tags/dataengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#dataengineering</a> <a href="https://data-folks.masto.host/tags/ai" class="mention hashtag" rel="nofollow noopener" target="_blank">#ai</a> <a href="https://data-folks.masto.host/tags/mlinproduction" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlinproduction</a>

Tim Kelloggthere’s a lot of really cool stuff in <a href="https://hachyderm.io/tags/MLEngineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#MLEngineering</a> that amounts to “train another model”. like using <a href="https://hachyderm.io/tags/SHAP" class="mention hashtag" rel="nofollow noopener" target="_blank">#SHAP</a> to automate feature selection (first you have to train a model though). or <a href="https://hachyderm.io/tags/ConceptSHAP" class="mention hashtag" rel="nofollow noopener" target="_blank">#ConceptSHAP</a> where you train simple linear models on the output of each neural net layer. or anomaly detection, or autoencoders, or...

Mikiko😳 My talk proposal to the <a href="https://data-folks.masto.host/tags/mlops" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlops</a> track was accepted to <a href="https://data-folks.masto.host/tags/DataCouncilAustin" class="mention hashtag" rel="nofollow noopener" target="_blank">#DataCouncilAustin</a> 2023 🤯 🎉 What an exciting way to start the year! 😃Looking forward to connecting with folks in Austin from March 28-30th on <a href="https://data-folks.masto.host/tags/mlops" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlops</a> <a href="https://data-folks.masto.host/tags/productionml" class="mention hashtag" rel="nofollow noopener" target="_blank">#productionml</a> <a href="https://data-folks.masto.host/tags/mlengineering" class="mention hashtag" rel="nofollow noopener" target="_blank">#mlengineering</a> <a href="https://data-folks.masto.host/tags/productiondatascience" class="mention hashtag" rel="nofollow noopener" target="_blank">#productiondatascience</a> <a href="https://data-folks.masto.host/tags/DataCouncilAustin2023" class="mention hashtag" rel="nofollow noopener" target="_blank">#DataCouncilAustin2023</a> <a href="https://data-folks.masto.host/tags/datacouncil" class="mention hashtag" rel="nofollow noopener" target="_blank">#datacouncil</a> Please feel free to connect with me on LI if you're attending or presenting! <a href="https://www.linkedin.com/in/mikikobazeley/" rel="nofollow noopener" target="_blank">https://www.linkedin.com/in/mikikobazeley/</a>

Recent searches

Search options

Administered by:

Server stats:

#mlengineering