<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Mihisara Nithadya on Medium]]></title>
        <description><![CDATA[Stories by Mihisara Nithadya on Medium]]></description>
        <link>https://medium.com/@nithadya?source=rss-85d0afd9ddf4------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*5l8q6mpSmuk5zpWjoCGqcA.jpeg</url>
            <title>Stories by Mihisara Nithadya on Medium</title>
            <link>https://medium.com/@nithadya?source=rss-85d0afd9ddf4------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 17:21:21 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@nithadya/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[MiniCPM-V 4.6: The FREE Tiny Vision Model Your Agents Need]]></title>
            <link>https://nithadya.medium.com/minicpm-v-4-6-the-free-tiny-vision-model-your-agents-need-1491ac0bfcb2?source=rss-85d0afd9ddf4------2</link>
            <guid isPermaLink="false">https://medium.com/p/1491ac0bfcb2</guid>
            <category><![CDATA[ai-agents-in-action]]></category>
            <category><![CDATA[vlm]]></category>
            <category><![CDATA[computer-vision]]></category>
            <category><![CDATA[opensource-ai]]></category>
            <category><![CDATA[minicpm]]></category>
            <dc:creator><![CDATA[Mihisara Nithadya]]></dc:creator>
            <pubDate>Wed, 20 May 2026 01:15:30 GMT</pubDate>
            <atom:updated>2026-05-20T01:15:30.477Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*W3DcfIG28cyTQfouFzQqiw.png" /></figure><p>MiniCPM-V 4.6 is not just another vision language model release. From a real product and system design perspective, it represents a shift in how multimodal AI can actually be deployed into working systems instead of staying limited to cloud based research setups.</p><p>The key idea behind this model is not maximum size, but maximum usability under real constraints such as cost, latency, hardware limits, and privacy. It is a <strong>1.3B parameter</strong> vision language model built using a <strong>SigLIP2 based vision encoder</strong> and a compact Qwen based language backbone. Even though the parameter size is relatively small compared to large cloud models, it still handles <strong>OCR</strong>, document reasoning, screenshot understanding, multi image analysis, video interpretation, and UI level visual grounding in a practical way.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-dDqLfARKCcbXgnuOgZGNw.png" /></figure><p>From an engineering perspective, one of the most important parts is how it handles visual data. Normally, vision models convert images and video frames into a large number of tokens, which increases memory usage and slows down inference. MiniCPM-V 4.6 reduces this problem using aggressive visual token compression and optimized encoding strategies, including mixed resolution compression techniques. This directly reduces computational cost while maintaining usable visual understanding quality.</p><p>Another important technical detail is the <strong>262K context window</strong>. This allows the model to process long sequences of visual and textual information together. In real applications, this is useful for analyzing multiple screenshots in sequence, long documents, or continuous video frames without losing context. This becomes important when building systems that need continuity instead of single step responses.</p><p>On the deployment side, the model is designed for practical execution environments rather than only high end cloud servers. It can run through frameworks such as llama.cpp, vLLM, SGLang, and Ollama, and supports quantized formats like GGUF and AWQ. This means it can be deployed on consumer GPUs, local servers, and in some cases even edge and mobile environments depending on optimization.</p><p>This changes the product economics significantly. Instead of relying only on paid API based inference for every visual task, systems can be designed to run locally or semi locally. This reduces per request cost, improves latency, and allows more control over data flow. For real product development, especially in startups or enterprise internal tools, this directly impacts scalability and margins.</p><p>One of the most valuable capabilities from a product point of view is GUI understanding. The model can interpret screenshots, recognize interface elements like buttons and input fields, and understand spatial layout. This allows a different type of automation where software can be operated visually instead of relying only on APIs or backend integration.</p><p>This is very important in real environments where many systems are still legacy based or lack proper integration layers. Instead of rebuilding systems, visual agents can be used to automate tasks directly on existing interfaces. This enables automation for workflows such as dashboard operations, form filling, data extraction, browser based tasks, and internal tool navigation.</p><p>This is where the concept of agentic systems becomes practical. Instead of simple chat based AI, this model can be part of systems that observe a screen, understand what is happening, decide the next action, and trigger automation tools to execute tasks. This creates a loop where perception and action are combined, which is the core idea behind modern AI agents.</p><p>Another important aspect is privacy and deployment flexibility. Since the model can run locally, sensitive visual data like documents, screenshots, and internal dashboards do not need to leave the system. This is important for industries where data control and compliance matter, such as finance, healthcare, logistics, and enterprise software environments.</p><p>From a technical efficiency perspective, the model also focuses on reducing inference overhead through optimized visual encoding pipelines. This improves throughput and reduces latency, which makes it more practical for real time applications compared to heavier multimodal systems.</p><p>Overall, MiniCPM-V 4.6 represents a shift in direction for multimodal AI systems. Instead of focusing only on scaling up model size, it focuses on making AI actually deployable in real environments with real constraints.</p><p>The real value is not just what the model can do in theory, but what it enables in production systems. It allows multimodal intelligence to move closer to devices, closer to users, and closer to real workflows instead of being locked inside expensive cloud infrastructure.</p><p>That is what makes it practically important for modern AI product development and system design.</p><h3>For More Details</h3><ul><li><strong>Official GitHub Repository:</strong> <a href="https://github.com/OpenBMB/MiniCPM-V">github.com</a></li><li><strong>Hugging Face Model Card:</strong> <a href="https://huggingface.co/openbmb/MiniCPM-V-4.6">huggingface.co</a></li><li><strong>MiniCPM-V-4.6-Thinking Model:</strong> <a href="https://huggingface.co/openbmb/MiniCPM-V-4.6-Thinking">huggingface.co</a></li><li><strong>Ollama Model Library:</strong> <a href="https://ollama.com/openbmb/minicpm-v4.6">ollama.com</a></li><li><strong>System Deployment CookBook:</strong> <a href="https://github.com/OpenSQZ/MiniCPM-V-CookBook">github.com</a></li><li><strong>vLLM Production Deployment Guide:</strong> <a href="https://github.com/OpenSQZ/MiniCPM-V-CookBook/blob/main/deployment/vllm/minicpm-v4_6_vllm.md">github.com</a></li><li><strong>llama.cpp Edge Optimization Guide:</strong> <a href="https://github.com/OpenSQZ/MiniCPM-V-CookBook/blob/main/deployment/llama.cpp/minicpm-v4_6_llamacpp.md">github.com</a></li><li><strong>Main Architecture Research Paper:</strong> <a href="https://arxiv.org/abs/2408.01800">arxiv.org</a></li><li><strong>Efficiency Strategy Research Paper:</strong> <a href="https://arxiv.org/abs/2509.18154">arxiv.org</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1491ac0bfcb2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Iceberg Effect Why the Sri Lanka Fuel QR System Failed at National Scale]]></title>
            <link>https://nithadya.medium.com/the-iceberg-effect-why-the-sri-lanka-fuel-qr-system-failed-at-national-scale-480b649878b0?source=rss-85d0afd9ddf4------2</link>
            <guid isPermaLink="false">https://medium.com/p/480b649878b0</guid>
            <category><![CDATA[kafka]]></category>
            <category><![CDATA[fuel-system]]></category>
            <category><![CDATA[solution-architecture]]></category>
            <category><![CDATA[system-design-concepts]]></category>
            <category><![CDATA[sri-lanka]]></category>
            <dc:creator><![CDATA[Mihisara Nithadya]]></dc:creator>
            <pubDate>Wed, 18 Mar 2026 18:41:53 GMT</pubDate>
            <atom:updated>2026-03-18T18:41:53.591Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*l0iI44cE_7zfF3Qg2KjeNg.png" /></figure><p>If you just look at it from the outside, building the Fuel QR system sounds like a weekend project for a few university interns. You just build a basic web app that generates a QR code, checks a vehicle number against a database, and cuts down a fuel quota when you scan it. On paper, the logic is super basic. It is just standard data entry with a clean user interface.</p><p>But when you take this deceptively simple solution and push it live to an entire country over a single night during an economic crisis, the reality hits you hard.</p><p>When you have millions of desperate people trying to get their daily fuel at the exact same time, you are no longer just coding an app. You are running mission critical national infrastructure. I have seen this exact movie before. The recent crash of the Fuel QR system was not because the code was bad. It was a massive failure in architectural strategy. It is a brutal lesson in what happens when you underestimate scale. There is a huge gap between just writing syntax and actually designing a system that survives extreme pressure.</p><h3>The illusion of Simplicity and The Iceberg Fallacy</h3><p>The main reason junior developers and the general public think massive projects are simple is because they confuse a simple user interface with a simple architecture. Because getting a QR code only takes a few clicks, everyone on social media claimed they could build the whole thing in a few hours. This is what we call the Iceberg Fallacy.</p><p>People only see the tip of the iceberg above the water. They do not see the massive structure hiding underneath.</p><p>When you compare what the user sees on their screen to what the backend actually has to do, the gap is terrifying.</p><ul><li>Typing in a vehicle number looks easy, but the backend has to make a Synchronous External API Call. That means it blocks everything else while it waits for a slow legacy database to answer.</li><li>Hitting submit requires massive Backpressure Management. The system has to handle millions of hits at once without crashing the old government servers it relies on.</li><li>Managing multiple vehicles requires Atomic Data Integrity. You have to handle one to many relationships perfectly without creating ghost records or accidentally deleting someone’s first car.</li><li>Scanning the QR at the shed requires Distributed State Synchronization and Graceful Degradation. The system needs to enforce quotas in real time across thousands of petrol sheds, even if the main database goes completely offline.</li></ul><h3>The Silent Killer Your Dependency is Your Ceiling</h3><p>Your system is only as strong as the weakest link it connects to. The Fuel QR app was tightly coupled to the Motor Traffic Department API to check vehicles. That API is an innocent legacy system that was never built to handle high speed traffic. Asking a slow database a few questions a day is fine. Blasting it with millions of requests all at once is basically an architectural assault.</p><p>Because they did not build a Service Layer or a Proxy Pattern to filter the traffic, the main system was completely naked. It was so unprotected that tech savvy guys were literally using Postman to bypass the broken website and hit the backend directly to register their cars.</p><p>As architects, we cannot magically make an old government database fast. But we can build protective layers around it using an Event Driven Architecture. Instead of hitting the old database directly, you use throttling and bulkheads. It works like this.</p><ol><li><strong>Submit</strong> The user submits a request and the system instantly says we got it.</li><li><strong>Queue</strong> That request goes into a durable messaging queue like Kafka or AWS SQS. Think of it as a secure waiting room.</li><li><strong>Process</strong> A background worker takes those requests one by one at a safe speed, maybe twenty a minute, so the old legacy system does not crash.</li><li><strong>Validate</strong> The actual validation happens safely without breaking anything.</li><li><strong>Generate and Notify</strong> A separate service generates the QR code and sends an SMS to the user a few hours later.</li></ol><h3>Data Evolution and The Amateur Truncate Trap</h3><p>When they relaunched the system years later, they had the typical mindset of just wiping the database and starting over. That is an amateur move. Trashing the database means millions of people have to register again. You waste money on SMS fees, burn server power, and make millions of people angry just to fix a few bad records.</p><p>The rollout failed because they made three terrible assumptions. They assumed everyone from 2023 was still alive. They assumed everyone still had the exact same recycled phone number. And they assumed nobody bought or sold a vehicle over the last three years.</p><p>In the real world, phone numbers get recycled every six months and thousands of cars get sold. Plus, they made a really stupid design choice. They restricted registration to a single phone number rather than using the chassis number as the ultimate unique key. You are fueling the vehicle, not the phone.</p><p>A smart architect would never wipe the whole database. They would use tactical purging. You talk to the motor traffic team and get a flat list of vehicles that were sold recently. You ask the phone companies for a list of dead SIM cards. Then you only delete those specific bad records. You clean the house in a few hours and keep the system working perfectly for the actual valid users.</p><h3>Strategic Rollouts Managing the Human Element</h3><p>The biggest mistake was launching it to the whole island on night one. When you do that, crashes are absolutely guaranteed. When a national app crashes, people panic on Facebook. That panic makes people refresh the page over and over again, which basically launches a self inflicted denial of service attack on your own servers.</p><p>A strong rollout needs to happen in phases so the team can monitor errors and fix things before everyone logs in. You have to roll out strategically.</p><ul><li><strong>Vehicle Type Release</strong> Open it for cars on Monday and bikes on Tuesday.</li><li><strong>Province Wise Release</strong> Start with the Western Province to see how much load the servers can take before opening it to the whole country.</li><li><strong>Age Based Release</strong> Let people who registered recently go first, then open it up to older accounts.</li></ul><h3>The Professional Mandate Why Engineers Must Say No</h3><p>Right after a system crashes, the first excuse you hear is that the politicians or the management forced them to launch early. This is exactly where we need to draw a hard line. Software engineering is a serious profession, just like being a doctor.</p><p>If a patient or even the president of the country demands the wrong medicine, a real doctor will never give it to them because the doctor is the professional. The software engineer is the professional here. As an architect, you need to have the guts to say no to a bad rollout plan. Your job is not to hit a random deadline. Your job is to protect the systems that the public relies on.</p><p>The main lesson here is that writing code is the easy part. Building systems that actually survive the real world is the hard part. It is time to stop thinking like a coder and start thinking like an architect.</p><h3>References and Further Reading</h3><ul><li><a href="https://www.google.com/search?q=https://youtu.be/iwygBhXy-wU"><strong>Fuel QR System Failure Explained</strong></a> by Krish Dinesh. A massive shoutout to this brilliant technical breakdown on YouTube. Much of the architectural analysis in this article, including the iceberg analogy, the Postman bypass issue, and the tactical database purging strategy, was heavily inspired by his deep dive into the rollout failure.</li><li><a href="https://martinfowler.com/articles/201701-event-driven.html"><strong>The Many Meanings of Event Driven Architecture</strong></a> by Martin Fowler. If you want to understand how to decouple systems and avoid the synchronous API trap that killed the legacy servers, this is the ultimate foundational guide.</li><li><a href="https://www.google.com/search?q=https://medium.com/%40saurabh.singh0829/backpressure-in-software-systems-147c2ed77bb5"><strong>Understanding Backpressure in Software Systems</strong></a>. A great concept to explore if you want to learn how to protect your legacy databases from being taken offline by your own frontend using traffic throttling.</li><li><a href="https://learn.microsoft.com/en-us/azure/architecture/patterns/circuit-breaker"><strong>Circuit Breaker Pattern for Distributed Systems</strong></a> by Microsoft. An excellent read on how to gracefully handle failed connections to external APIs without crashing your entire application.</li><li><a href="https://aws.amazon.com/message-queue/"><strong>Introduction to Message Queuing</strong></a> by AWS. A straightforward breakdown of how tools like SQS and Kafka act as shock absorbers during massive traffic spikes to prevent system overload.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=480b649878b0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The End of UI-First SaaS: How Agentic AI and API-First Products Are Taking Over]]></title>
            <link>https://nithadya.medium.com/the-end-of-ui-first-saas-how-agentic-ai-and-api-first-products-are-taking-over-b006951ac86d?source=rss-85d0afd9ddf4------2</link>
            <guid isPermaLink="false">https://medium.com/p/b006951ac86d</guid>
            <category><![CDATA[api]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[saas]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Mihisara Nithadya]]></dc:creator>
            <pubDate>Wed, 21 Jan 2026 13:43:48 GMT</pubDate>
            <atom:updated>2026-02-04T13:18:10.291Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZzJBKqPH3Ftn_2iGUYBSnQ.jpeg" /></figure><p>For over a decade, the mantra in the tech world was simple: build a beautiful app, make it easy to use, put a big “Sign Up” button on the front page, and the users will flock to your product. This “UIfirst” (User Interface) approach was the golden rule of software. We obsessed over pixelperfect layouts, color palettes that evoked “trust,” and the psychological placement of every notification bell.</p><p>But standing here in 2026, I can tell you as an AI Engineer and Startup CEO: The rules didn’t just change the game was replaced entirely.</p><p>We are witnessing The Great Decoupling. In plain English, this means the value of software is being separated from the screen you look at. If you’re still building software just for human eyes to look at, you’re building a museum, not a business. The “wow factor” has moved from the surface to the core.</p><p><strong>The Old Paradigm: UI/UX and the Mirage of Engagement</strong></p><p>The traditional approach to software products (SaaS) was centered around human psychology. The logic was linear:</p><ol><li>Attractive UI/UX: A visually stunning interface that is easy to navigate to reduce human friction.</li><li>Engagement Hooks: Calls to action (CTAs), gamification, and notifications designed to keep a human “time-on-site” high.</li><li>Conversion Funnel: Encourage free trials and retain users through the “experience” of using the tool.</li></ol><p>This worked well when the human sitting behind the screen was the primary actor. Every click mattered. Every color choice could influence subscription numbers. Companies invested months sometimes years perfecting their dashboards, believing this would guarantee adoption. We were essentially building “digital workspaces” for humans to labor in.</p><p><strong>The New Era: When “Users” Are Robots (Agentic AI)</strong></p><p>Fast forward to 2026, and the landscape has shifted. Users are no longer the only, or even the primary, actors in the digital ecosystem. Enter Agentic AI systems and Autonomous Automation Agents.</p><p>Recent industry data shows that over 40% of big business applications now use taskspecific AI agents. These are smart programs capable of acting on behalf of users:</p><ol><li>They make purchases, process data, and manage workflows automatically.</li><li>They integrate directly with APIs, bypassing web interfaces entirely.</li><li>They do not care about your shiny dashboards, modern UI components, or CTAs.</li></ol><p>These agents only care about functionality and data accessibility. The “wow factor” of your interface is invisible to these autonomous systems. Even the most beautifully designed web app is irrelevant if your product cannot “plug” into AI agents efficiently. You might have the best storefront in the world, but your customers are now drones that only care about the loading dock in the back.</p><p><strong>Why Traditional SaaS Approaches Are Failing</strong></p><p>Many SaaS products are still built around the old paradigm, chasing metrics that no longer move the needle in an agent-driven economy.</p><ol><li>The Ghost User: AI assistants don’t watch your 5minute tutorial videos. If your software requires a human to “figure it out” via a UI, an AI agent will simply choose a different tool that speaks its language.</li><li>The Death of the Button: We used to obsess over the “Submit” button. But an AI agent doesn’t “click” buttons; it executes programmed actions via JSON payloads. Your onboarding flows and popups are now invisible friction.</li><li>Invisible Software: Many products are becoming “invisible infrastructure.” If you’re still measuring success by “clicks” or “scrolls,” you are measuring a ghost.</li></ol><p><strong>The Future: API-first, Agent-Ready Products</strong></p><p>The shift is toward Agent-Ready Architectures AI and API-first designs created to work seamlessly with autonomous systems. This requires a fundamental “Level Up” in how we build and monetize.</p><ol><li><strong>Talk Like a Machine: The “USB-C” Moment (MCP)</strong></li></ol><p>Software now uses the Model Context Protocol (MCP). Think of it as the “USB-C” for AI. It allows any AI agent to plug into your software and understand exactly what it does, what data it has, and what actions it can take instantly. If you don’t have an MCPcompliant “plug,” the AI won’t use your tool. Your API documentation is no longer for human developers; it is the “training manual” for the AI itself.</p><p><strong>2. Solving the Authentication Paradox: AgentNative Auth</strong></p><p>Traditional SaaS uses MFA (Multi-Factor Authentication) and Captchas to keep bots out. In an agentic world, blocking the bot is blocking your customer.</p><p>Companies that cling to high-friction, legacy security gates are effectively de-platforming themselves from the agent economy. We are moving toward Agent-native Auth:</p><ul><li>Programmable API Keys: Digital keys that grant smart bots secure access.</li><li>Granular Permissions: Letting an agent “Read Data” but not “Change Bank Details.”</li><li>Bounded Autonomy: Security that allows an agent to act, but within strict cryptographic boundaries.</li></ul><p><strong>The Pricing Pivot: From “Seats” to “Outcomes”</strong></p><p>We used to charge $50 per “seat” (per person). But why charge for a seat when one AI agent can do the work of a dozen people? The world is moving to Outcome-Based Pricing.</p><ul><li>Monetize via Usage: Charging for data processed or API calls made.</li><li>Monetize via Results: Charging per successful task completion. Instead of waiting for human users to subscribe, products will sell themselves through autonomous ecosystems.</li></ul><p><strong>Does the Screen Still Matter? (Trust over Tasks)</strong></p><p>UI/UX still matters for humans, but its job has fundamentally changed. In 2026, the UI is no longer about Tasks; it is about Trust. You don’t log in to do the work; you log in to check the work.</p><ul><li>The Supervisory Pane: The screen is now a “Command Center.” You log in once a day to see an audit log a report of what your AI agents did while you were sleeping. The focus is now on transparency and proving the AI isn’t hallucinating.</li><li>The Emergency Brake (Human-on-the-Loop): If an AI agent wants to perform a highstakes action (like spending $10,000), your software should “pause” and ask for human permission. This is Human-on-the-loop (HOTL) governance where the UI serves as a high-stakes checkpoint.</li></ul><p><strong>The Bottom Line for Founders and Developers</strong></p><p>If you are building software in 2026 and beyond:</p><ol><li>Stop obsessing over dashboards as your primary growth lever. The “Brain” (API) comes first; the “Face” (UI) comes second.</li><li>Focus on APIs, MCP integration, and automation-first design. Make your product discoverable and usable by AI agents, and the revenue will follow.</li><li>Sell the result, not the subscription. Outcome-based value is the only way to scale in an economy where work is done by machines.</li><li>The era of UI/UX dominance is giving way to the era of agentic software. Those who keep building for eyeballs instead of agents will find themselves with a beautiful product that nobody and nothing uses.</li></ol><p><strong>References &amp; Further Reading</strong></p><ol><li>Rōnin Consulting (Jan 2026): The Rise of AI Agents: Impact on SaaS.</li><li>Zenit Data &amp; Varun Mayya (Jan 2026): The AI Shift in SaaS is a Business Model Reset.</li><li>BCG (Oct 2025): How Agentic AI is Transforming Enterprise Platforms.</li><li>Deloitte Insights (Nov 2025): SaaS Meets AI Agents: Federated Workflow Services.</li><li>The New Stack (Nov 2025): Building APIs for AI, Not Just For Developers.</li><li>Jakob Nielsen (Sept 2025): The End of UI: How AI Agents Will Replace Human Interaction.</li></ol><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b006951ac86d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Clickbait, Lies, and Hidden Fees: How ‘Dark Patterns’ Trick You Online]]></title>
            <link>https://nithadya.medium.com/clickbait-lies-and-hidden-fees-how-dark-patterns-trick-you-online-4e7bd9b0cf01?source=rss-85d0afd9ddf4------2</link>
            <guid isPermaLink="false">https://medium.com/p/4e7bd9b0cf01</guid>
            <category><![CDATA[ui-design]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[ui-ux-design]]></category>
            <category><![CDATA[ux-research]]></category>
            <dc:creator><![CDATA[Mihisara Nithadya]]></dc:creator>
            <pubDate>Mon, 29 Sep 2025 20:22:54 GMT</pubDate>
            <atom:updated>2025-09-29T20:41:29.738Z</atom:updated>
            <content:encoded><![CDATA[<p>A simple guide to understanding the deceptive designs that manipulate us into clicking, buying, and sharing things we don’t want to.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZrCLBIoOFBfuZuJY9eoxjg.png" /><figcaption>figure 0: Topic</figcaption></figure><h4><strong>What Are Dark Patterns, Really?</strong></h4><p>You sign up for a “free” trial, and a month later, your bank statement shows a mysterious charge. You try to buy a plane ticket, and the price suddenly jumps up on the final payment page.</p><p>Sound familiar?</p><p>These frustrating moments aren’t accidents or bugs. They are the result of something called a <strong>dark pattern, </strong>a user interface that has been carefully designed to trick you into doing something you didn’t intend to.</p><p>Think of them as digital booby traps, set by companies to boost their sales or sign-ups at your expense. They work by exploiting our basic human psychology, and once you learn to see them, you’ll start noticing them everywhere.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/926/1*_p4v4agajWeOldXXBtcKvA.png" /><figcaption>Figure 1: What is Dark Patterns</figcaption></figure><h3>The Deceiver’s Playbook: 9 Common Digital Traps</h3><p>Dark patterns come in many shapes and sizes, but a few pop up more than others. Here are some of the most common ones you’ve probably encountered.</p><h4><strong><em>The Most Wanted Dark Patterns</em></strong></h4><ul><li><strong>🏨 The Roach Motel:</strong> It’s incredibly easy to sign up, but when you want to leave, it’s a nightmare. <strong>Example:</strong> Amazon was sued by the FTC for allegedly trapping users in its Prime subscription with a confusing, multi-step cancellation process.</li><li><strong>💸 Hidden Costs:</strong> The price you see at first isn’t the price you pay. Extra fees (like service fees or taxes) are only revealed at the very last step. <strong>Example:</strong> Ticketmaster and other ticketing sites are famous for this.</li><li><strong>😥 Confirmshaming:</strong> The website uses manipulative language to make you feel guilty for declining an offer, like a pop-up that says, “No thanks, I prefer to pay full price.”</li><li><strong>🛒 Sneak into Basket:</strong> An extra item gets added to your shopping cart without you actively choosing it, often through a pre-checked box.</li><li><strong>🤫 Privacy Zuckering:</strong> Named after Mark Zuckerberg, this trick fools you into sharing more personal data than you intended. <strong>Example:</strong> LinkedIn was sued for accessing users’ address books and spamming their contacts.</li><li><strong>🔄 Forced Continuity:</strong> You sign up for a “free” trial with your credit card, and when it ends, you are automatically and silently charged.</li><li><strong>🎭 Disguised Ads:</strong> An advertisement is designed to look like a normal piece of content or a navigation button, tricking you into clicking on it.</li><li><strong>🎣 Bait-and-Switch:</strong> You try to do one thing, but a different, undesirable thing happens instead. <strong>Example:</strong> Microsoft famously made the “X” close button on its Windows 10 upgrade pop-up <em>start</em> the upgrade instead of closing the window.</li><li><strong>🛑 Forced Action:</strong> You’re forced to do something you don’t want to in order to do something you do. <strong>Example:</strong> Many e-commerce sites force you to create an account to make a purchase, instead of offering a guest checkout.</li></ul><h3>Why These Tricks Work: The Psychology Behind the Deception</h3><p>These patterns aren’t just bad design; they’re smart design used for the wrong reasons. They are effective because they prey on basic human tendencies.</p><ul><li><strong>We’re Lazy (Default Bias):</strong> We tend to stick with pre-set options. That’s why boxes for “sign me up for marketing emails” are often pre-checked.</li><li><strong>We Hate Losing (Loss Aversion):</strong> We feel the pain of a loss more than the joy of a gain. “Confirmshaming” frames declining an offer as a loss, making us uncomfortable.</li><li><strong>We Follow Through (Sunk Cost Fallacy):</strong> Once we’ve started a process (like filling out forms), we feel committed and are less likely to back out when we discover hidden fees later.</li><li><strong>We Trust the Crowd (Social Proof):</strong> We assume if others are doing something, it must be right. This is why some sites use fake testimonials or pop-ups saying “Jane from New York just bought this!”</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/825/1*6leaOOOc1utEKEUaSBd2MA.png" /><figcaption>Figure 2: Good UX vs. Dark Patterns -A Question of Intent</figcaption></figure><p>Good design makes your goal easier. Dark patterns make the business’s goal easier, even if it hurts you.</p><h3>More Than Just Annoying: The Real Cost of Deception</h3><p>A company might trick a user into one extra sale, but the long-term cost is huge. When people feel deceived, they lose trust. And trust is the foundation of any good relationship, in business or otherwise.</p><p>In today’s world, a bad experience doesn’t stay quiet. It gets shared on social media, leading to a damaged reputation that can be hard to fix. Laws are also catching up. Companies like <strong>Intuit (TurboTax)</strong> have faced massive lawsuits a-<strong>$141 million settlement-</strong>for using these deceptive tactics.</p><p>Ethical design isn’t just about doing the right thing; it’s the smart thing to do for long-term success.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ac1xt03UcI_64rqw83_HTw.png" /><figcaption>Figure 3: Cost of Deceit</figcaption></figure><h3>Don’t Get Fooled: How You Can Fight Back</h3><p>While companies are responsible for designing honestly, we aren’t powerless. Here are a few simple things you can do to protect yourself.</p><p><strong><em>Your Shield Against Dark Patterns</em></strong></p><p><strong>Slow Down:</strong> Dark patterns often rely on you being in a hurry. Take an extra moment to read everything on the page before you click, especially during checkout.</p><p><strong>Question Everything:</strong> If an offer seems too good to be true, or if the language feels confusing or pushy, pause. Be skeptical of countdown timers and “limited stock” claims.</p><p><strong>Uncheck the Boxes:</strong> Always look for pre-checked boxes that sign you up for newsletters, add-ons, or data sharing you don’t want.</p><p><strong>Vote with Your Wallet:</strong> Whenever you can, choose to support businesses that are transparent and treat you with respect.</p><h3>Building a Better, More Honest Web</h3><p>Dark patterns are a betrayal of the user. They turn the internet into a landscape of traps instead of a space of tools.</p><p>The responsibility to fix this lies with the designers and companies who build our digital world. But by learning to spot these tricks and choosing to support ethical businesses, we can all contribute to a more honest, transparent, and user-friendly web for everyone.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lWOb-cCqBd7FQzso_7Gzvw.png" /><figcaption>Figure 4: For Designers &amp; Businesses: A Quick Ethical Checklist</figcaption></figure><h4>References</h4><ol><li><em>Dark pattern — Wikipedia</em>. (2025). en.wikipedia.org.</li><li><em>10 Common Dark Patterns in UX and How to Avoid Them</em>. (2025). Net Solutions.</li><li><em>Are Dark Patterns Illegal in 2025? Honda, the Law, and UX Loopholes</em>. (2025). Ketch.</li><li><em>How website dark patterns can cost you</em>. (2025). CHOICE.</li><li><em>Types — Confirmshaming — Deceptive Patterns</em>. (2025). deceptive.design.</li><li><em>Chapter 20: Forced action — Deceptive Patterns</em>. (2025). deceptive.design.</li><li><em>FTC vs. Intuit (TurboTax) Settlement Information</em>. (2025). Federal Trade Commission.</li></ol><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4e7bd9b0cf01" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[END-TO-END WEATHER DATA PIPELINE]]></title>
            <link>https://nithadya.medium.com/end-to-end-weather-data-pipeline-8c146d87b51c?source=rss-85d0afd9ddf4------2</link>
            <guid isPermaLink="false">https://medium.com/p/8c146d87b51c</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[design]]></category>
            <category><![CDATA[nlp]]></category>
            <dc:creator><![CDATA[Mihisara Nithadya]]></dc:creator>
            <pubDate>Sat, 27 Sep 2025 01:29:44 GMT</pubDate>
            <atom:updated>2025-09-27T19:19:10.225Z</atom:updated>
            <content:encoded><![CDATA[<p>END-TO-END WEATHER DATA PIPELINE</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/1*bWLV1gKtu7ca3RxugVvA1Q.jpeg" /></figure><p>In today’s data-driven world, the ability to design and run efficient data pipelines is one of the most valuable skills for aspiring data engineers and data scientists. Recently, I came across an interesting project that demonstrates how to build a simple but powerful weather data pipeline using only open-source technologies.</p><p>🚀 What does the pipeline do?</p><p>The goal of this project is to fetch real-time weather data from an external API, transform it into a clean and usable format, and load it into a database for analysis. All of this is automated and containerized, making it scalable and reproducible.</p><p>🛠️ Tools Used</p><ul><li>Docker → Containerization for consistent environments</li><li>Apache Airflow → Workflow orchestration and scheduling</li><li>Python → Extracting raw data from the WeatherStack API</li><li>PostgreSQL → Storing both raw and transformed weather data</li><li>dbt (Data Build Tool) → Transformations and modeling inside the database</li></ul><p>⚙️ How it works (ETL Flow)</p><p>• Extract: Python scripts pull raw weather data from the WeatherStack API.</p><p>• Load: Raw data is loaded into PostgreSQL.</p><p>• Transform: dbt models transform the raw data into structured, analysis-ready tables.</p><p>• Orchestrate: Apache Airflow schedules and manages the pipeline tasks, ensuring smooth execution.</p><p>• Containerize: Docker makes the entire solution portable and easy to deploy anywhere.</p><p>🎓 Why is this important for Data Science Undergraduates?</p><p>For undergraduates studying Data Science, this kind of project is extremely valuable because it bridges the gap between theory and real-world application:</p><p>• Beyond Machine Learning: Many students focus only on modeling, but in industry, 80% of the work is data preparation and pipeline building. This project teaches you how to handle that critical part.</p><p>• Connecting Tools Together: You learn how multiple open-source tools (Airflow, dbt, Docker, PostgreSQL) integrate to solve a practical problem.</p><p>• Real-Time Data Handling: Working with live APIs (like weather data) gives hands-on experience with real-world, messy, and constantly updating datasets.</p><p>• Portfolio-Ready Project: Having an end-to-end data pipeline project in your portfolio shows recruiters that you understand not just analysis, but also the engineering side of data science.</p><p>• Career Preparation: These are the same tools used by data engineers and data scientists in companies, so early exposure gives you a huge career advantage.</p><p>💡 Takeaway: For data science undergraduates, building projects like this is a great way to move beyond classroom assignments and gain practical, industry-ready experience in ETL, data pipelines, and modern data stack tools.</p><p><strong><em>GitHub Link: </em></strong><a href="https://github.com/nithadya/Weather-API-Pipeline-Airflow.git"><em>Weather-API-Pipeline</em></a><em>.</em></p><p><strong><em>Credits:</em></strong><em> </em><a href="https://github.com/stevetran77">Steve Tran<em>.</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8c146d87b51c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ Demystifying Your First React Project: The Complete Guide to Understanding Every File and Folder]]></title>
            <link>https://nithadya.medium.com/demystifying-your-first-react-project-the-complete-guide-to-understanding-every-file-and-folder-dd9c4d57301d?source=rss-85d0afd9ddf4------2</link>
            <guid isPermaLink="false">https://medium.com/p/dd9c4d57301d</guid>
            <dc:creator><![CDATA[Mihisara Nithadya]]></dc:creator>
            <pubDate>Fri, 13 Jun 2025 12:12:43 GMT</pubDate>
            <atom:updated>2025-06-13T12:37:35.482Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*O9r2wF4uC9sOPsHA2l5eRw.png" /></figure><p><strong><em>Master your Vite + React + TypeScript project structure from the ground up</em></strong></p><h3>🎯 What You’ll Master</h3><p>Starting a new React project can feel overwhelming when you’re faced with dozens of files and folders. Don’t worry! by the end of this comprehensive guide, you’ll understand exactly what each piece does and why it’s there. We’ll transform that confusing file tree into a clear roadmap for building amazing React applications.</p><p><strong>You’ll Learn:</strong></p><ul><li>The purpose of every single file and folder</li><li>How different parts of your project communicate with each other</li><li>When and why you should modify specific files</li><li>Professional patterns used in real-world React development</li><li>Best practices for organizing your growing codebase</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ToAmsCvy9ZNH_pC_lLgo_Q.png" /></figure><h3>1. 📁 The Complete Project Anatomy</h3><p>When you create a new React project with Vite and TypeScript, you get a well-organized structure that might look intimidating at first. Let’s break it down into digestible pieces:</p><pre>📦 my-react-app/<br>├── 📁 node_modules/           # Your project&#39;s dependencies<br>├── 📁 public/                 # Static files served directly<br>├── 📁 src/                    # Your application source code<br>├── 📄 .gitignore              # Files Git should ignore<br>├── 📄 eslint.config.js        # Code quality rules<br>├── 📄 index.html              # HTML template<br>├── 📄 package-lock.json       # Dependency version lock<br>├── 📄 package.json            # Project configuration<br>├── 📄 README.md               # Project documentation<br>├── 📄 tsconfig.app.json       # TypeScript app settings<br>├── 📄 tsconfig.json           # Main TypeScript config<br>├── 📄 tsconfig.node.json      # Node.js TypeScript config<br>└── 📄 vite.config.ts          # Build tool configuration</pre><p>Think of this structure like a well-organized house: each room has a specific purpose, and everything has its place. The root level contains your project’s “foundation” files, while the src folder is where you’ll spend most of your time building features.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zLz2Q-OQvj5lN8IjZYowig.png" /></figure><h3><strong>2. 🏠 Root Level Files: Your Project’s Foundation</strong></h3><h3>📄 package.json - Project’s Identity Card</h3><p>This file is like your project’s passport — it contains all the essential information about your application. When someone (including yourself six months from now) looks at your project, this file tells them everything they need to know.</p><p><strong>What’s Inside:</strong></p><ul><li><strong>Project Metadata: </strong>Name, version, description, and author information</li><li><strong>Scripts</strong>: Commands you can run like npm run dev to start development or npm run build to create a production version</li><li><strong>Dependencies</strong>: External libraries your project needs to function (like React itself)</li><li><strong>Development Dependencies</strong>: Tools needed only during development (like TypeScript and testing libraries)</li></ul><p><strong>Key Sections Explained:</strong> The scripts section is your command center. When you run npm run dev, npm looks here to find what command to execute. The dependencies section lists libraries that will be included in your final application, while devDependencies contains tools that help you build and develop but won&#39;t be part of the final product.</p><p><strong>When You’ll Edit This:</strong> Every time you install a new package with npm install, this file updates automatically. You might manually edit it to add custom scripts or modify project metadata.</p><h3>📄 vite.config.ts -Your Build System Control Panel</h3><p>Vite is your project’s build tool — think of it as a sophisticated assembly line that takes your source code and transforms it into something browsers can understand efficiently. This configuration file tells Vite exactly how to do its job.</p><p><strong>What It Controls:</strong></p><ul><li><strong>Development Server</strong>: How your local development environment behaves</li><li><strong>Build Process</strong>: How your code gets optimized for production</li><li><strong>Plugin Integration</strong>: Additional tools that extend Vite’s capabilities</li><li><strong>Path Resolution</strong>: How your project resolves import statements</li></ul><p><strong>Why It Matters:</strong> Vite is incredibly fast because it uses modern JavaScript features and intelligent caching. During development, it serves your files almost instantly. When building for production, it optimizes everything for the best performance.</p><p><strong>When You’ll Modify It:</strong> You’ll typically edit this when adding new plugins (like adding support for a CSS framework), configuring path aliases to make imports cleaner, or customizing the build process for specific deployment needs.</p><h3>📄 TypeScript Configuration Files -Your Code Quality Guardians</h3><p>TypeScript configuration is split across multiple files for better organization:</p><p><strong>tsconfig.json</strong> acts as the main coordinator, referencing other configuration files. It’s like a manager that delegates specific tasks to specialized team members.</p><p><strong>tsconfig.app.json</strong> contains settings specifically for your application code. It defines things like which JavaScript features you can use, how strict the type checking should be, and what files should be included in compilation.</p><p><strong>tsconfig.node.json</strong> handles configuration for Node.js tools like your Vite config file. This separation ensures that different parts of your project can have appropriate settings.</p><p><strong>Why Multiple Files?</strong> This separation allows different parts of your project to have different TypeScript settings. Your application code might need browser-specific types, while your build tools need Node.js types.</p><p><strong>What They Control:</strong></p><ul><li>Type checking strictness (catching errors before they happen)</li><li>Which JavaScript features you can use</li><li>How modules are resolved and imported</li><li>Path mapping for cleaner imports</li></ul><h3>📄 eslint.config.js -Your Code Quality Coach</h3><p>ESLint is like having a experienced developer looking over your shoulder, pointing out potential issues and suggesting improvements. It catches common mistakes, enforces consistent coding style, and helps prevent bugs.</p><p><strong>What It Does:</strong></p><ul><li><strong>Error Prevention</strong>: Catches potential bugs before they reach production</li><li><strong>Style Consistency</strong>: Ensures your code follows consistent patterns</li><li><strong>Best Practice Enforcement</strong>: Guides you toward proven patterns</li><li><strong>React-Specific Rules</strong>: Understands React patterns and common pitfalls</li></ul><p><strong>How It Helps:</strong> ESLint runs continuously as you code, underlining potential issues in your editor. It’s not just about finding errors — it teaches you better coding practices over time.</p><p><strong>When You’ll Configure It:</strong> You might adjust rules based on your team’s preferences, add new rule sets for specific libraries you’re using, or customize error severity levels.</p><h3>📄 index.html - Your Application’s Front Door</h3><p>This HTML file is where everything begins. It’s the first thing browsers load, and it provides the foundation for your React application to build upon.</p><p><strong>Key Elements:</strong></p><ul><li><strong>The Root Element</strong>: The &lt;div id=&quot;root&quot;&gt;&lt;/div&gt; is where React will mount your entire application</li><li><strong>Script Tag</strong>: Points to your main TypeScript file to start the React application</li><li><strong>Meta Tags</strong>: Provide information for browsers, search engines, and social media</li><li><strong>External Resources</strong>: Links to fonts, favicons, and other assets</li></ul><p><strong>Why It’s Special:</strong> Unlike traditional websites where you might have multiple HTML files, React applications typically use just this one. React then dynamically creates and manages all the content you see.</p><p><strong>When You’ll Edit It:</strong> You’ll modify this when adding meta tags for SEO, linking to external fonts or stylesheets, or adding analytics scripts.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LbSeq6B9FFHaLcRcXgCIVA.png" /></figure><h3>3. 📁 The src Directory -Where the Magic Happens</h3><p>The src folder is your creative workspace — it’s where you’ll spend 90% of your time building features, components, and functionality.</p><h3>📄 main.tsx-Your Application’s Ignition Key</h3><p>This file starts your entire React application. It’s the bridge between your HTML template and your React components.</p><p><strong>What Happens Here:</strong></p><ol><li><strong>Import Essential Libraries</strong>: Brings in React, ReactDOM, and your main App component</li><li><strong>Find the Root Element</strong>: Locates the div with id=”root” in your HTML</li><li><strong>Create a React Root</strong>: Establishes React’s control over that element</li><li><strong>Render Your App</strong>: Mounts your entire application to the DOM</li></ol><p><strong>React.StrictMode Explained:</strong> This wrapper component doesn’t render anything visible but enables additional development checks. It helps catch common mistakes and warns about deprecated features. Think of it as training wheels that help you write better React code.</p><p><strong>When You’ll Modify It:</strong> Rarely, but you might add global providers for state management, authentication context, or theme providers that need to wrap your entire application.</p><h3>📄 App.tsx-Your Application’s Headquarters</h3><p>App.tsx is typically your application’s main hub — the root component that orchestrates everything else. It’s where you define your overall application structure and manage top-level concerns.</p><p><strong>Common Responsibilities:</strong></p><ul><li><strong>Layout Structure</strong>: Defining headers, navigation, main content areas, and footers</li><li><strong>Routing</strong>: Determining which components to show based on the current URL</li><li><strong>Global State</strong>: Managing application-wide data like user authentication status</li><li><strong>Theme Management</strong>: Handling dark/light mode switching</li><li><strong>Error Boundaries</strong>: Catching and handling unexpected errors gracefully</li></ul><p><strong>Why It’s Important:</strong> App.tsx sets the tone for your entire application. It’s where you establish patterns that will be used throughout your project. A well-structured App component makes it easier to maintain and expand your application over time.</p><p><strong>As Your Project Grows:</strong> Initially, App.tsx might be simple, but as your application becomes more complex, you’ll likely refactor it to delegate responsibilities to specialized components and services.</p><h3>📁 components/-Your UI Building Block Library</h3><p>The components folder is where you’ll build the reusable pieces of your user interface. Think of components like LEGO blocks — individual pieces that you can combine in different ways to build complex interfaces.</p><p><strong>Organization Strategies:</strong></p><p><strong>By Function:</strong></p><ul><li>common/ - Components used across multiple features (buttons, inputs, modals)</li><li>layout/ - Structural components (headers, sidebars, footers)</li><li>features/ - Components specific to particular features (user profile, dashboard)</li></ul><p><strong>By Complexity:</strong></p><ul><li>ui/ - Basic, presentational components with minimal logic</li><li>containers/ - Components that manage data and business logic</li><li>pages/ - Top-level components that represent entire views</li></ul><p><strong>Component Best Practices:</strong> Each component should have a single, clear responsibility. A good component is easy to understand, test, and reuse. Components should be like helpful team members — reliable, predictable, and good at their specific job.</p><p><strong>Professional Tips:</strong></p><ul><li>Keep components small and focused</li><li>Use TypeScript interfaces to define props clearly</li><li>Include proper error handling and loading states</li><li>Write components that are easy to test</li><li>Consider accessibility from the beginning</li></ul><h3>📁 hooks/ — Your Custom Logic Library</h3><p>Custom hooks are where you extract and reuse stateful logic. They’re like specialized tools that you can use across different components.</p><p><strong>Why Custom Hooks Matter:</strong> Instead of copying and pasting the same logic between components, you can create a custom hook that encapsulates that behavior. This makes your code more maintainable and testable.</p><p><strong>Common Custom Hook Patterns:</strong></p><ul><li><strong>Data Fetching</strong>: Hooks that handle API calls and manage loading/error states</li><li><strong>Local Storage</strong>: Hooks that sync component state with browser storage</li><li><strong>Form Handling</strong>: Hooks that manage form state and validation</li><li><strong>Authentication</strong>: Hooks that manage user login state and permissions</li></ul><p><strong>When to Create Custom Hooks:</strong> If you find yourself writing similar useEffect or useState logic in multiple components, that’s a perfect candidate for a custom hook.</p><h3>📁 services/ — Your External Communication Center</h3><p>The services folder contains code that handles communication with external systems — APIs, databases, third-party services, and more.</p><p><strong>Why Separate Services:</strong> By keeping API calls and external integrations in dedicated service files, you create a clear separation between your UI logic and data access logic. This makes your code easier to test, maintain, and modify.</p><p><strong>Common Service Patterns:</strong></p><ul><li><strong>API Services</strong>: Functions that handle HTTP requests to your backend</li><li><strong>Authentication Services</strong>: Logic for logging in, out, and managing tokens</li><li><strong>Data Transformation</strong>: Functions that convert data between different formats</li><li><strong>Caching</strong>: Logic for storing and retrieving frequently used data</li></ul><p><strong>Benefits of This Approach:</strong> When your API changes, you only need to update the service files, not every component that uses that data. It also makes it easier to add features like caching, error handling, and request retries.</p><h3>📁 types/ — Your TypeScript Contract Library</h3><p>The types folder contains TypeScript type definitions that describe the shape of your data throughout the application.</p><p><strong>Why Centralize Types:</strong> Having all your types in one place ensures consistency across your application. When you change a type definition, TypeScript will help you find all the places that need to be updated.</p><p><strong>Common Type Categories:</strong></p><ul><li><strong>Entity Types</strong>: Descriptions of your main data objects (User, Product, Order)</li><li><strong>API Types</strong>: Types for request and response data</li><li><strong>Component Props</strong>: Types for component properties</li><li><strong>Utility Types</strong>: Helper types that are used across multiple areas</li></ul><p><strong>TypeScript Benefits:</strong> Good type definitions catch errors at compile time instead of runtime, provide excellent autocomplete in your editor, and serve as documentation for other developers.</p><h3>📁 utils/ — Your Helper Function Toolkit</h3><p>The utils folder contains pure functions that perform common tasks throughout your application.</p><p><strong>What Goes Here:</strong></p><ul><li><strong>Data Formatting</strong>: Functions to format dates, currency, or text</li><li><strong>Validation</strong>: Functions to check if data meets certain criteria</li><li><strong>Calculations</strong>: Mathematical operations or data transformations</li><li><strong>Constants</strong>: Application-wide constant values</li></ul><p><strong>Why Separate Utilities:</strong> Utility functions are often needed in multiple places. By centralizing them, you avoid code duplication and create a consistent way to handle common tasks.</p><p><strong>Best Practices:</strong> Utility functions should be pure — they should not have side effects or depend on external state. This makes them predictable and easy to test.</p><h3>📁 assets/ — Your Media and Resource Collection</h3><p>The assets folder contains files that get processed by your build tool — images, fonts, icons, and other media that are imported directly into your components.</p><p><strong>Assets vs Public Folder:</strong> Files in assets get processed by Vite — they can be optimized, renamed, and included in your bundle. Files in the public folder are served directly without processing.</p><p><strong>When to Use Assets:</strong> Use the assets folder for images and files that are imported into your components, logos that might change based on themes, or resources that benefit from build-time optimization.</p><h3>📄 CSS Files — Your Application’s Visual Foundation</h3><p><strong>index.css</strong> contains global styles that apply to your entire application. This is where you define CSS variables, reset styles, and utility classes that are used throughout your project.</p><p><strong>App.css</strong> contains styles specific to your App component. As your application grows, you’ll likely have CSS files that correspond to specific components or features.</p><p><strong>CSS Organization Strategies:</strong></p><ul><li>Use CSS custom properties (variables) for consistent theming</li><li>Implement a CSS reset to ensure consistent rendering across browsers</li><li>Create utility classes for common styling patterns</li><li>Consider CSS modules or styled-components for component-scoped styles</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3T03ivEy7FYf688GsAebYw.png" /></figure><h3>4. 📁 The public/ Directory — Your Static Asset Vault</h3><p>The public folder contains files that are served directly to users without any processing. These files maintain their original names and can be accessed via their URL path.</p><h3>📄 favicon.ico — Your Brand’s Tiny Ambassador</h3><p>The favicon is the small icon that appears in browser tabs, bookmarks, and when users add your site to their home screen. It’s a small detail that contributes to your application’s professional appearance.</p><h3>📄 manifest.json — Your Progressive Web App Configuration</h3><p>The manifest file transforms your React application into a Progressive Web App (PWA). It defines how your app appears when installed on mobile devices and desktop computers.</p><p><strong>What It Enables:</strong></p><ul><li>Installation prompts on mobile devices</li><li>Custom app icons and splash screens</li><li>Offline functionality configuration</li><li>Native app-like experience</li></ul><h3>📄 robots.txt — Your Search Engine Instructions</h3><p>This file tells search engine crawlers which parts of your site they should or shouldn’t index. It’s an important file for SEO and controlling how search engines interact with your application.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9XjHK9-a59B_ibMOVbYS1w.png" /></figure><h3>5. 🔄 How Everything Works Together</h3><p>Understanding individual files is important, but seeing how they interact gives you the complete picture:</p><p><strong>The Development Flow:</strong></p><ol><li>You run npm run dev which executes the script defined in package.json</li><li>Vite reads its configuration and starts the development server</li><li>The browser loads index.html, which includes the script tag pointing to main.tsx</li><li>main.tsx imports and renders your App component</li><li>App.tsx orchestrates your application’s layout and routing</li><li>Individual components handle specific UI pieces</li><li>Services manage external data</li><li>TypeScript ensures type safety throughout</li><li>ESLint provides code quality feedback</li></ol><p><strong>The Build Process:</strong> When you run npm run build, Vite transforms your development code into optimized files ready for production. It bundles JavaScript, processes CSS, optimizes images, and creates a dist folder with everything needed to deploy your application.</p><h3>6. 🎯 Professional Development Patterns</h3><h3>Component Organization Evolution</h3><p><strong>Beginner Approach:</strong> Start with simple folders like components, hooks, and utils. This works well for small projects and helps you understand the basics.</p><p><strong>Intermediate Approach:</strong> Organize by feature rather than file type. Group related components, hooks, services, and types together based on what part of the application they serve.</p><p><strong>Advanced Approach:</strong> Implement domain-driven design where code is organized around business concepts rather than technical concerns. This scales well for large applications with multiple teams.</p><h3>File Naming Conventions</h3><p><strong>Components</strong>: Use PascalCase (UserProfile.tsx, ShoppingCart.tsx) <strong>Hooks</strong>: Use camelCase starting with “use” (useLocalStorage.ts, useAuth.ts) <strong>Utilities</strong>: Use camelCase (formatDate.ts, validateEmail.ts) <strong>Types</strong>: Use PascalCase often matching the entity they describe (User.ts, Product.ts)</p><h3>Import Organization</h3><p>Organize your imports in a consistent order:</p><ol><li>External libraries (React, third-party packages)</li><li>Internal utilities and services</li><li>Type imports</li><li>Relative imports (components in the same feature)</li><li>CSS imports</li></ol><h3>7. 🚀 Next Steps for Your React Journey</h3><h3>Immediate Actions</h3><ol><li><strong>Explore Your Project</strong>: Open each file and read through the code, even if you don’t understand everything yet</li><li><strong>Make Small Changes</strong>: Modify text in App.tsx and see the results</li><li><strong>Create Your First Component</strong>: Build a simple component in the components folder</li><li><strong>Experiment with Styles</strong>: Try changing values in the CSS files</li></ol><h3>Building Your Skills</h3><ol><li><strong>Start Small</strong>: Build components with simple functionality first</li><li><strong>Read Documentation</strong>: The React and TypeScript docs are excellent resources</li><li><strong>Practice Patterns</strong>: Try implementing common patterns like forms, lists, and modals</li><li><strong>Learn Gradually</strong>: Don’t try to master everything at once</li></ol><h3>Professional Development</h3><ol><li><strong>Version Control</strong>: Use Git to track your changes and experiment safely</li><li><strong>Testing</strong>: Learn to write tests for your components</li><li><strong>Performance</strong>: Understand how to optimize React applications</li><li><strong>Accessibility</strong>: Build applications that work for everyone</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Rn5JwuMNMblwb3Ibv1IhXQ.png" /></figure><h3>8. 💡 Pro Tips for React Success</h3><h3>Development Workflow</h3><ul><li>Use the React Developer Tools browser extension for debugging</li><li>Set up your editor with React and TypeScript extensions</li><li>Learn keyboard shortcuts for your development environment</li><li>Use the browser’s developer tools to inspect and debug</li></ul><h3>Code Quality</h3><ul><li>Write descriptive component and function names</li><li>Add comments for complex logic</li><li>Keep components small and focused</li><li>Use TypeScript to catch errors early</li></ul><h3>Performance Considerations</h3><ul><li>Understand when components re-render</li><li>Learn about React’s built-in optimization tools</li><li>Optimize images and assets</li><li>Consider lazy loading for large components</li></ul><h3>Troubleshooting Common Issues</h3><ul><li><strong>White Screen</strong>: Check the browser console for JavaScript errors</li><li><strong>Components Not Updating</strong>: Verify state is being set correctly</li><li><strong>Import Errors</strong>: Check file paths and export/import syntax</li><li><strong>Styling Issues</strong>: Inspect elements to see which CSS rules are applied</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JinK948a7nL559_fRqFT-w.png" /></figure><h3>🎉 You’re Ready to Build Amazing Things!</h3><p>Understanding your React project structure is like having a detailed map of a new city. You now know where everything lives, why it’s there, and how different parts work together to create your application.</p><p><strong>Remember:</strong></p><ul><li>Every expert was once a beginner</li><li>Making mistakes is part of learning</li><li>The React community is welcoming and helpful</li><li>Practice is more valuable than perfection</li></ul><p><strong>Your Next Challenge:</strong> Pick one component in your project and try to understand exactly what it does. Then, make a small modification and see what happens. This hands-on exploration will solidify your understanding faster than any tutorial.</p><p>The journey of mastering React is exciting and rewarding. You have all the foundational knowledge you need to start building. Now it’s time to code, experiment, and create something amazing!</p><p><em>Ready to dive deeper? Start building your first feature and remember — every line of code you write makes you a better developer. Happy coding! 🚀</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dd9c4d57301d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Embark on Your AI Journey]]></title>
            <link>https://nithadya.medium.com/embark-on-your-ai-journey-9e1fca9f7daa?source=rss-85d0afd9ddf4------2</link>
            <guid isPermaLink="false">https://medium.com/p/9e1fca9f7daa</guid>
            <category><![CDATA[ai-for-beginners]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[beginners-guide]]></category>
            <dc:creator><![CDATA[Mihisara Nithadya]]></dc:creator>
            <pubDate>Tue, 06 Aug 2024 14:43:44 GMT</pubDate>
            <atom:updated>2024-08-06T14:43:44.775Z</atom:updated>
            <content:encoded><![CDATA[<p><strong><em>“Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow”</em></strong></p><p>In the ever-evolving landscape of technology, the realms of AI and machine learning shine brightly, beckoning curious minds to explore their depths. For those ready to embark on this transformative journey, there is a guidebook that stands out: <em>“Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow”</em> by Aurélien Géron.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*w2Oo3hM-I6pAypiIoRWBCQ.jpeg" /></figure><h4><strong>The Gateway to AI Mastery</strong></h4><p>Imagine yourself as a brave adventurer, stepping into the vast, uncharted territories of artificial intelligence. Your mission is to harness the power of data, decode complex algorithms, and create intelligent systems. But every great adventure needs a map, and this book is your ultimate guide.</p><h4>Why This Book is Your Essential AI Companion</h4><p>Navigating the world of AI can be overwhelming, but <em>“Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow”</em> simplifies the journey with its practical and hands-on approach. Here’s why this book is an invaluable resource for beginners and how it helps transform them into experts:</p><ul><li><strong>Hands-On Learning</strong>: Dive into coding and experiment with real-world data from the start. The book guides you through practical examples, making complex concepts more accessible and engaging.</li><li><strong>Comprehensive Coverage</strong>: This book covers a broad spectrum of topics, including:</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0PWx1cQjDndhTFiqP05rJg.png" /></figure><ul><li><strong>Real-World Applications</strong>: Each chapter includes exercises that help you apply machine learning algorithms to solve real-world problems. This practical experience is crucial for building a solid foundation in AI.</li><li><strong>Accessible Code Examples</strong>: Written in Python, one of the most popular programming languages for data science, the book uses Scikit-Learn, Keras, and TensorFlow to demonstrate various machine learning techniques. The clear and concise code examples make it easy for beginners to follow along.</li><li><strong>Clear Explanations</strong>: Aurélien Géron has a talent for breaking down complex ideas into simple, understandable terms. This makes the book an excellent starting point for anyone new to AI and machine learning.</li><li><strong>Progression from Beginner to Expert</strong>: The book is structured to guide readers from basic concepts to advanced techniques. It starts with fundamental machine learning principles, gradually introduces more complex models and deep learning frameworks, and eventually dives into state-of-the-art techniques. This progression ensures that as you work through the book, your skills evolve from beginner to expert.</li></ul><h4>Access the Code and Resources</h4><p>To complement the book, you can access all the code examples and preparation details in the official repository. This GitHub repository includes:</p><ul><li>Complete code examples from the book</li><li>Data preparation scripts</li><li>Jupyter notebooks for hands-on practice</li></ul><p>You can explore and download the resources here: <a href="https://github.com/nithadya/Hands-On-ML.git"><em>Hands-On ML GitHub Repository</em></a></p><h4>How to Maximize Your Learning</h4><p>To get the most out of this book, consider these tips:</p><ol><li><strong>Code Along with the Examples</strong>: As you read, type out the code examples and run them on your own computer. This hands-on practice reinforces your learning and helps you understand the material better.</li><li><strong>Experiment with Your Own Projects</strong>: Don’t just follow the examples in the book. Apply what you’ve learned to your own datasets and projects. This will help you gain confidence and develop problem-solving skills.</li><li><strong>Engage with the Community</strong>: Join online forums, study groups, or social media communities dedicated to AI and machine learning. Connecting with other learners and practitioners can provide support, answer questions, and offer different perspectives.</li></ol><blockquote>Conclusion</blockquote><blockquote><em>“Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow”</em> is more than just a book; it’s a comprehensive guide that bridges the gap between theory and practice. Whether you’re just starting your journey into AI and machine learning or looking to deepen your knowledge, this resource is an excellent investment in your education.</blockquote><blockquote>From foundational concepts to advanced techniques, this book will guide you every step of the way, helping you grow from a beginner into an AI expert. Embark on your AI adventure with confidence, and let this book be your trusted companion. Happy learning, and may your journey be filled with discovery and innovation!</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9e1fca9f7daa" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>