Writing

Content creation with artificial intelligence models is booming since Nov. 2022. If many architectures exist, most of them belong to a handful of with interesting properties. Explore the diverse landscape of Generative AI architectures with this taxonomy, detailing four primary families: Variational Autoencoders (VAE), Generative Diffusion Models (GDM), Generative Adversarial Networks (GANs), and Transformers.

In this article, I provide strategies for effectively communicating complex technical concepts to non-technical stakeholders, particularly in machine learning contexts. I emphasize the importance of thorough preparation, clear and concise explanations, and the use of relatable analogies to bridge the gap between technical and non-technical perspectives. By adopting these approaches, you can enhance understanding and foster productive collaborations with stakeholders across various domains.

A quick post on applying a divide & conquer approach to download large files in network constrained environment. In this article, I address the challenges of downloading large files over unreliable networks by introducing a practical three-step approach: splitting the large file into smaller chunks using the UNIX split command, transferring these manageable segments individually, and then recombining them on the client side with the cat command. This method ensures successful file retrieval even in environments with intermittent connectivity.

How to edit raw tracks into a compelling audio story? Transform your raw podcast recordings into compelling audio stories with essential editing techniques. In this article, I share practical insights on using Digital Audio Workstations (DAWs) like Adobe Audition, GarageBand, Logic Pro, or Audacity to enhance your podcast's quality. Learn how to address common recording issues and apply fundamental editing tools to captivate your audience effectively.

Multiprocessing is hard, here’s a practical example to be on the right track in 5mins. In this article we explore the differences between multithreading and multiprocessing and their impact on performance on a OCR use case. I provide practical guidance on implementing concurrent processing in Python using gRPC, enabling efficient parallelism and improved performance in your applications.

A comprehensive guide to application profiling in Python, covering tools and techniques for identifying performance bottlenecks. In this article, I explore systematic profiling techniques to pinpoint areas where your code may be slowing down, enabling you to make informed improvements that enhance overall performance.

A checklist to help you deliver content faster. In this article, I share three essential strategies to enhance your podcast editing workflow: mastering your digital audio workstation (DAW) shortcuts, planning your narrative structure before editing, and utilizing templates and effect presets to streamline the process.

Model inference at the edge with Raspberry Pi and the Coral USB edge TPU. In this article, I demonstrate how to develop an end-to-end mask detection system using TensorFlow and Kafka, integrating IoT devices equipped with cameras and AI capabilities to identify mask compliance in real-time. By deploying Python-based face detection models at the edge and utilizing Kafka for asynchronous communication, the system efficiently collects and processes data from various locations, enhancing public safety measures.

Watch out for append() behavior. In this article, I explore a common pitfall when implementing backtracking algorithms in Go, particularly concerning the behavior of the append() function with slices. I demonstrate how unintended side effects can arise when append() modifies underlying slice data during recursive backtracking, leading to incorrect results.

A quick and practical walkthrough for creating a Telegram bot using the Go programming language. Learn to register your bot, handle requests, set up a webhook, and deploy your bot, all illustrated with a working example, the RapGeniusBot, which generates creative rap lyrics on command.

All the things I wish I knew before I started doing a Domain Model. I describe a step-by-step guide on conducting Domain Modeling Interviews to effectively acquire domain expertise from subject- matter experts. By following this approach, you can develop interpersonal skills that complement your engineering expertise, leading to the creation of robust domain models that accurately represent the problem space.

Launching a machine learning project doesn't have to break the bank. In this article, I share practical strategies to minimize costs without compromising performance. Learn how to leverage serverless computing, optimize resource allocation, and make informed decisions that keep your project both efficient and budget-friendly.

An in-depth exploration of Apache NiFi, the open-source tool that simplifies and automates dataflow management across diverse systems. Nifi is designed to automate the flow of data between systems. I explore its core components—such as FlowFiles, Processors, and Connections—and demonstrate how NiFi's user-friendly, flow-based programming interface simplifies the creation and management of complex data pipelines. By leveraging NiFi's capabilities, you can efficiently route, transform, and manage data across diverse sources and destinations, enhancing your data workflow's reliability and scalability.

Introducing a streamlined method for deploying & updating machine learning models using TensorFlow Serving. Using a "sidecar" pattern that employs a model poller container alongside the TensorFlow Serving container to automate the detection and loading of new model versions from a storage bucket.

A step-by-step tutorial demonstrating how to deploy ML models in production using TensorFlow Serving and Kubernetes, with practical examples using object detection models and Google Cloud Platform.

A comprehensive guide on building a complete ML pipeline, from lyrics data collection to production deployment, using a text generation model (RNN) as a case study. Covers data scraping, model training on AWS, and production deployment with an Apache HTTP Server and Flask.