I'm a Machine Learning Researcher and Engineer. Here I write posts on the intersection of deep learning theory and efficient machine learninghttp://localhost:4321/en-ushttp://localhost:4321/rethinking-quantization-aware-training-why-your-qat-length-is-probably-wrong/http://localhost:4321/rethinking-quantization-aware-training-why-your-qat-length-is-probably-wrong/Training quantized neural networks involves a fundamental trade-off: how should you divide your compute budget between full-precision pretraining and quantization-aware training?Thu, 30 Oct 2025 20:37:42 GMThttp://localhost:4321/understanding-flash-attention-writing-the-algorithm-from-scratch-in-triton/http://localhost:4321/understanding-flash-attention-writing-the-algorithm-from-scratch-in-triton/Why is Flash Attention so fast? Find out how Flash Attention works. Afterward, we'll polish our understanding by writing a GPU kernel of the algorithm in Triton.Sun, 12 Jan 2025 18:36:55 GMThttp://localhost:4321/speed-up-pytorch-with-custom-kernels-but-it-gets-progressively-darker/http://localhost:4321/speed-up-pytorch-with-custom-kernels-but-it-gets-progressively-darker/It's all about making your models run faster, from flicking a magic “compile” switch to writing your own custom GPU code. In each step, we’ll implement an innocent softmax function, but things are about to get dark by the end.Sat, 04 Jan 2025 00:34:24 GMThttp://localhost:4321/simple-ways-to-speedup-your-pytorch-model-training/http://localhost:4321/simple-ways-to-speedup-your-pytorch-model-training/If all machine learning engineers want one thing, it's faster model training — maybe after good test metrics.Tue, 28 May 2024 20:16:11 GMThttp://localhost:4321/swift-actors-common-problems-and-tips/http://localhost:4321/swift-actors-common-problems-and-tips/Swift actors are a powerful tool to address data races and make your code thread-safe. However, it is also quite a sophisticated concept that requires deep understanding to write efficient and bug-free code. Tue, 13 Jun 2023 12:32:57 GMThttp://localhost:4321/i-contributed-to-pytorch-heres-what-i-learned/http://localhost:4321/i-contributed-to-pytorch-heres-what-i-learned/When you see something that does not work in an omnipresent framework, you believe it can't be completely broken, right?Mon, 20 Mar 2023 16:23:35 GMThttp://localhost:4321/conquer-data-races-with-swift-actors/http://localhost:4321/conquer-data-races-with-swift-actors/Unleash the power of Swift concurrency with Actors! Get all the information you need in this comprehensive articleTue, 07 Feb 2023 19:08:18 GMThttp://localhost:4321/dive-into-swifts-memory-management/http://localhost:4321/dive-into-swifts-memory-management/Swift uses ARC to track and deallocate unused objects. Learn about the three types of reference counts and how ARC works — in this detailed post.Sun, 08 Jan 2023 19:33:03 GMThttp://localhost:4321/data-binding-in-swiftui-tips-tricks-and-best-practices/http://localhost:4321/data-binding-in-swiftui-tips-tricks-and-best-practices/Want to create dynamic and responsive user interfaces in SwiftUI? Data binding is the key! In this tutorial, I'll show you how to use @State, @ObservedObject, @EnvironmentObject, and @Binding to keep your user interface in sync with your dataFri, 30 Dec 2022 13:16:07 GMThttp://localhost:4321/ios-app-as-a-microservice-using-swiftui-in-modular-app/http://localhost:4321/ios-app-as-a-microservice-using-swiftui-in-modular-app/The modular architecture is excellent. But how to implement it effectively with SwiftUI? From its core, SwiftUI is state-driven, and it can be tricky to modularize an app and define exact responsibility borders. Wed, 19 Oct 2022 13:00:23 GMThttp://localhost:4321/ios-app-as-a-microservice-modularize-your-app-with-tuist/http://localhost:4321/ios-app-as-a-microservice-modularize-your-app-with-tuist/This is the second article in a series on modular app architecture. In this post, I will cover implementation details using TuistFri, 07 Oct 2022 10:14:04 GMThttp://localhost:4321/ios-app-as-a-microservice-build-robust-app-architecture/http://localhost:4321/ios-app-as-a-microservice-build-robust-app-architecture/What will you choose: MVVM, MVC, VIPER? Those all are local and problem-specific architectures. But how to structure your app on a larger scale to make it scalable and well-organized?Fri, 16 Sep 2022 07:43:25 GMThttp://localhost:4321/exploring-swiftui-layout-protocol-creating-custom-layout/http://localhost:4321/exploring-swiftui-layout-protocol-creating-custom-layout/Apple introduces new SwiftUI's Layout protocol with the release of iOS 16. It is a powerful tool for constructing custom views with SwiftUI elegance.Thu, 11 Aug 2022 21:00:18 GMThttp://localhost:4321/swiftui-navigation-is-a-mess-heres-what-you-can-do/http://localhost:4321/swiftui-navigation-is-a-mess-heres-what-you-can-do/Managing navigation in pure SwiftUI is hard and leads to messy solutions. In this post, I will show you how you can manage views effectivelySat, 30 Jul 2022 17:55:14 GMThttp://localhost:4321/suffix-automaton-and-rickroll/http://localhost:4321/suffix-automaton-and-rickroll/Easy to understand explanation of suffix automaton with implementation. Finally, generating correct Rickroll lyrics suffix automatonSun, 17 Jul 2022 13:57:17 GMThttp://localhost:4321/using-threads-in-swift/http://localhost:4321/using-threads-in-swift/Swift provides DispatchQueue as an excellent layer above raw threads. But sometimes you want to use a low-level thread APIFri, 13 May 2022 08:44:49 GMThttp://localhost:4321/swiftui-advanced-animation/http://localhost:4321/swiftui-advanced-animation/I'm going to show how complex SwiftUI views can be animated efficiently using VectorArithmetic protocol with Accelerate library for fast computations.Thu, 05 May 2022 04:00:00 GMThttp://localhost:4321/swift-async-algorithms-module/http://localhost:4321/swift-async-algorithms-module/Apple released the first version of the async swift algorithms package. It provides tools and algorithms to use with the introduced not that far ago Async SequenceWed, 27 Apr 2022 20:21:00 GMThttp://localhost:4321/treap-algorithm-explained/http://localhost:4321/treap-algorithm-explained/Binary search trees are mostly hard. Writing red-black tree is a nightmare. Here, I'm going to explain one of the easiest, yet efficient and powerful balanced binary tree — treap or cartesian treeMon, 25 Apr 2022 04:00:00 GMThttp://localhost:4321/swift-type-placeholders/http://localhost:4321/swift-type-placeholders/Type placeholders were recently introduced in Swift 5.6. Get in touch with new useful Swift feature.Thu, 21 Apr 2022 04:00:00 GMThttp://localhost:4321/quick-guide-to-async-await-in-swift/http://localhost:4321/quick-guide-to-async-await-in-swift/Everything you need to know about new Swift asynchronous features. Async await, main actor, task, async get, and possible use cases — all covered.Sat, 16 Apr 2022 08:50:00 GMThttp://localhost:4321/top-7-subtle-swift-features/http://localhost:4321/top-7-subtle-swift-features/Here, I collected Swift features that are less known and can be useful when you prepare for interviews or want to deepen your Swift knowledge.Sat, 09 Apr 2022 10:41:05 GMThttp://localhost:4321/note-taking-apps/http://localhost:4321/note-taking-apps/Here I cover note-taking apps for productivuty and creating your own knowledge databaseWed, 18 Aug 2021 16:51:00 GMThttp://localhost:4321/mystery-of-mach-o-object-file-builders/http://localhost:4321/mystery-of-mach-o-object-file-builders/I’m going to tell you about the internals of the Mach-O file and give an introduction to the simple relocatable object file structureThu, 29 Apr 2021 19:59:09 GMThttp://localhost:4321/skip-list-indexation-and-kth-maximum/http://localhost:4321/skip-list-indexation-and-kth-maximum/Skip List is a nice structure that lets you to perform insertions, searches, and finding n-th maximum. In this post I focus on skip list indexationThu, 05 Nov 2020 22:49:32 GMThttp://localhost:4321/how-deep-neural-networks-train/http://localhost:4321/how-deep-neural-networks-train/Here, I combine the explanation of Neural Nets with coding. By the end, we will develop a basic neural network and try to solve usual problemsFri, 08 May 2020 02:25:40 GMT