Posts

Showing posts with the label Security

Book Review: Software Security for Developers

Image
With the growing reliance on AI-assisted coding, managing application security is more critical than ever, as humans ultimately have to own the code they are responsible for. As the authors of Software Security for Developers note early on: "While developers often focus on libraries, frameworks, and tools at the mid-level, true security stems from foundational knowledge of standards, protocols, and patterns, as well as adherence to corporate and industry security practices." The first section provides the big picture and does a good job of explaining the high-level security vocabulary of modern software systems, which is further unraveled in subsequent chapters. The following sections move into deep cryptography primitives (AES, RSA, ECC), implementing secure transport layers (mTLS, X.509 certificates), and mastering enterprise-grade identity patterns like OAuth2, OpenID Connect (OIDC), PKCE, WebAuthn, and microservice call-chain authorization. The code examples are Java-cen...

This Week I Learned - Week #7 2025

This Week I Learned -  * Phi-4 is a 14B parameter state-of-the-art small language model (SLM) that excels at complex reasoning and conventional language processing. Phi-4 is available in GitHub Models. * You can now attach images and work with them directly in GitHub Copilot Chat. Share screenshots of errors and Copilot will interpret the image and resolve the issue.  * A group led by Elon Musk offered $97.4 billion to buy the nonprofit controlling OpenAI, aiming to prevent the AI startup from becoming a for-profit company. Musk co-founded OpenAI with Sam Altman in 2015 but left before it became successful. He later started a competing AI company, xAI, in 2023. OpenAI is trying to transition to a for-profit model to secure more funding for AI development. Musk sued Altman and others in August 2023, arguing they prioritized profit over public good. Musk believes OpenAI should return to its original mission of being open-source and focused on safety. * The gitingest service...

This Week I Learned - Week #48 2024

Image
This Week I Learned -   * "India continues to remain a super important market for Meta globally. It’s one of the largest user markets for us and the trends that we’re seeing here are definitely about continued growth and engagement across all our platforms. It’s not just Instagram, Facebook, and WhatsApp, but even for the recently launched Threads, India is one of the largest markets globally with over 200 million downloads. We’re seeing continued momentum and growth, across the board. Also, we are witnessing an equally strong adoption of newer products such as AI platform Llama. India provides the largest user base for Meta’s platforms. I think we are almost sort of using India as the playbook, you know, for global. Consider this: India was one of the first countries to test and launch (Instagram) Reels in 2020. Today, it’s one of the strongest products that we have globally. India is also the first country where we launched Payments on WhatsApp Business. Also, India is the b...

This Week I Learned - Week #39 2024

Image
This Week I Learned -  *  Cursor is a fork of VS Code . As a standalone application (and not an extension), Cursor has more control over the UI of the editor, enabling greater AI integration. Regardless of whether you use the free, pro or business version of Cursor , all generated code is yours and free to be used however you like, including commercially.  * Llama 3.2 is Meta’s newest release of vision and text models that fit on edge devices, enabling developers to build more private and personalized AI experiences. It includes small and medium-sized vision LLMs (11B and 90B) and lightweight, text-only models (1B and 3B) that support image reasoning and on-device use cases. * OpenAI's new family of AI models titled o1-preview are being trained via reinforcement learning and 'chain of thought' in order to answer queries. o1-preview is optimized for solving math and science problems, writing code, and solving word problems, with its expert reasoning and analysis. It excels...

This Week I Learned - Week #30 2024

Image
This Week I Learned -  * Microsoft's recent estimates show that the Crowdstrike outage impacted 8.5 million devices. The issue stemmed from a Crowdstrike cybersecurity software update that included a configuration change. This faulty configuration file stopped Windows machines from starting up. Although the outage was not triggered by a cybersecurity breach, malicious actors exploited the situation. Numerous phishing campaigns, masquerading as IT support or CrowdStrike, sent out harmful links or solicited funds to fix the issue. This event starkly illustrates the critical need for delivering high-quality software. * Modern artificial intelligence (AI) systems are powered by foundation models.  Llama 3 is a herd of language models  [PDF] that natively support multilinguality, coding, reasoning, and tool usage.  * Mistral Large 2 is not entirely open source like Llama, since developers are required to acquire a license for commercial use of Mistral's model. * The ...

This Week I Learned - Week #23 2024

Image
This Week I Learned -  * Microsoft 365 Education is a productivity cloud suite. Microsoft 365 builds on top of Office 365 apps such as Word, Excel, PowerPoint, and Outlook and provides enhanced management, security, and compliance tools. Microsoft 365 Education has three tiers of academic plans: A1, A3, and A5. A5 has the most comprehensive plan features [ PDF ]. Microsoft partnered with experts to develop Learning Accelerators tools Reading Progress and Reading Coach. *  DBRX is Databricks' Foundation Model * In Forrester's 21-criterion evaluation of AI Foundation Models for Language providers, Google, Databricks, NVIDIA emerged as Leaders in Q2 2024. * The first car was driven by a neural network in 1988. Twenty-five years later, Andrej Karpathy took his first demo ride in a Waymo. A decade after that, the company received its driverless permit. That’s thirty-five years of rigorous  engineering, testing, refinement, and regulatory navigation to go from p...

This Week I Learned - Week #21 2024

Image
This Week I Learned -  * Azure VMware Solution is available in in 33 Azure regions globally including India Central.  *  Power Automate now brings AI innovation to robotic process automation (RPA) by giving users the option to train the desktop recorder in a more natural and flexible way. Multimodal AI recording for desktop flows uses screen understanding and speech processing to let users teach the desktop recorder how to automate a process. AI flows leverage the simplicity of natural language and the power of rapidly evolving large language model (LLM) abilities to create more intelligent, flexible, and dynamic automation. Instead of building out complex rules-based processes from beginning to end, users work with AI to create an automation plan. When the plan is executed, the LLM considers all the resources available to it and dynamically selects the most appropriate actions to achieve the desired outcome.  * There are separate Microsoft 365 offerings for ho...

This Week I Learned - Week #7 2024

Image
This Week I Learned -  *  clipboard.js is a lightweight JavaScript library (3KB gzipped) that simplifies copying text to the clipboard without relying on Flash or any bulky frameworks. It's a handy tool for web developers who want to enhance user experience when dealing with clipboard operations.  Part of a full drawing from  Generative AI in a Nutshell - how to survive and thrive in the age of AI * Aya is a new open-source, state-of-the-art, massively multilingual, generative large language model (LLM) and dataset from Cohere. This model built by 3,000+ independent researchers both outperforms existing open-source models and covers more than double the number of languages of existing models – 101 different languages. The model significantly expands coverage to more than 50 previously under-supported languages.  *  O’Reilly has created a GenAI policy for its experts to follow when leveraging GenAI in the creation of their own works. It expects authors an...

What is Red Teaming Generative AI?

Image
Paraphrased from the HBR article  How to Red Team a Gen AI Model  - A red team is a group that pretends to be an enemy, attempts a physical or digital intrusion against an organization at the direction of that organization, then reports back so that the organization can improve their defenses. Red teams work for the organization or are hired by the organization.  It is a process of testing your cybersecurity effectiveness through the removal of defender bias by applying an adversarial lens to your organization. The term was popularized during the Cold War when the U.S. Defense Department tasked “red teams” with acting as the Soviet adversary, while blue teams were tasked with acting as the United States or its allies.  In October 2023, Biden's AI executive order mandated "red teaming" - a structured hacking effort to uncover flaws in certain high-risk generative models. Red teaming generative AI is specifically designed to generate harmful content that has no clear ...

This Week I Learned - Week #35 2023

Image
This Week I Learned - * In terms of speed and rendering quality SumatraPDF ranges at the top among free PDF-viewers since it has changed its library basis to MuPDF. PyMuPDF provides access to many important functions of MuPDF from within a Python environment.  * Ghostscript was first released in 1991 as a freely available open source (GPL) alternative to Adobe’s PostScript. Ghostscript became highly sought after by OEMs who wanted to license it for their own products. Artifex was created in 1993 to facilitate licensing of Ghostscript under commercial licensing terms. *  The latest IMDb data is available on-demand through a GraphQL-backed API via AWS Data Exchange . * Azure Form Recognizer is now Azure AI Document Intelligence . Azure Document Intelligence now supports Aadhaar and PAN card. Document Intelligence Identity document (ID) model combines Optical Character Recognition (OCR) with deep learning models to analyze and extract key information from id...

This Week I Learned - Week #31 2023

Image
This Week I Learned -  * Magenta is an open source research project exploring the role of machine learning as a tool in the creative process. This Tone Transfer demo link shows how to generate audio and turn your voice into a musical instrument of your choice.   * Joshua Lochner (original creator of Transformers.js), now at Hugging Face, has released a legendary demo that can transcribe your audio recordings super fast right in the browser powered by WebRTC and Web ML tech. - Whisper Web *  Gorilla is an  open-source , Apache 2.0 licensed (use commercially without any obligations) LLM tailored to serve API calls for your stated tasks. It is trained on three massive machine learning hub datasets: Torch Hub, TensorFlow Hub and HuggingFace.  With Gorilla CLI , you can simply state your objective, and the tool will generate potential commands for over 1500 APIs, including Kubernetes, AWS, GCP, Azure, and many more.  * The Gunicorn "Green Unicorn" ...

This Week I Learned - Week #26 2023

Image
This Week I Learned -  * A data warehouse processes structured data, while a data lake processes structured, semi-structured, unstructured, and raw binary data. * To shift security left means to implement security measures during the entire development lifecycle, rather than at the end of the cycle. * East-west traffic , in a networking context, is the transfer of data packets from server to server within a data center. The term east-west for this type of traffic comes from network diagram drawings that usually depict local area network (LAN) traffic horizontally.  * East/west traffic is traffic flowing around the Kubernetes cluster, service-to-service or service-to-datastore. *  Tailwind CSS is an open source CSS framework. The main feature of this library is that, unlike other CSS frameworks like Bootstrap, it does not provide a series of predefined classes for elements such as buttons or tables. Tailwind is designed for use with components like React, Angular, Vue, et...