<?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 Evgeni Gomziakov on Medium]]></title>
        <description><![CDATA[Stories by Evgeni Gomziakov on Medium]]></description>
        <link>https://medium.com/@gomzkov?source=rss-28ee7e3e3626------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*nWQHskfeLox6Kglvr7IWEQ.jpeg</url>
            <title>Stories by Evgeni Gomziakov on Medium</title>
            <link>https://medium.com/@gomzkov?source=rss-28ee7e3e3626------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 00:22:24 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@gomzkov/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[Reduce Token Waste]]></title>
            <link>https://gomzkov.medium.com/reduce-wasted-tokens-d36eb5b46442?source=rss-28ee7e3e3626------2</link>
            <guid isPermaLink="false">https://medium.com/p/d36eb5b46442</guid>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[ai-tools]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <category><![CDATA[software-development]]></category>
            <dc:creator><![CDATA[Evgeni Gomziakov]]></dc:creator>
            <pubDate>Thu, 25 Jun 2026 05:13:55 GMT</pubDate>
            <atom:updated>2026-09-11T15:46:37.595Z</atom:updated>
            <content:encoded><![CDATA[<h4>Practical ways to make AI coding agents cheaper and faster with RTK, Headroom, CodeGraph, Ponytail, and Caveman.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AyvLpVOT55Xjn2q4QBzjLQ.png" /></figure><p>AI coding agents are getting expensive.</p><p>The obvious fix is a cheaper model. Cutting wasted tokens works better.</p><p>Agents waste context on: terminal logs, full files, repeated grep loops, old state, verbose replies, and code that did not need to be written.</p><p>It also hurts quality, not only cost. More irrelevant context means slower runs, more retries, and a higher chance the agent loses the task, misses the right file, or changes the wrong thing.</p><p>I looked at a few tools trying to reduce this from different angles:</p><h3>RTK</h3><ul><li>Best when the terminal is the problem.</li><li>It intercepts shell commands and compresses their output before it reaches the model: git, tests, builds, package managers, Docker, grep, file reads, and more.</li><li>Reported savings: 60–90% on supported commands. In their own example, a 30 minute session drops from 118k tokens to 24k.</li><li>Limit: it works on shell commands only. The agent’s built in read and grep tools don’t pass through it, so file reads aren’t compressed unless the agent uses the shell.</li></ul><h3>Headroom</h3><ul><li>Best when context comes from many messy sources.</li><li>A broader compression layer. It compresses tool outputs, logs, files, RAG chunks, and conversation history before they reach the model. It is also reversible: the original is cached, so the agent can pull it back if it actually needs it.</li><li>Worth noting: Headroom uses RTK under the hood for shell output, so I would not count them as two fully separate layers.</li><li>Its own benchmarks report 47–92% savings, depending on the workflow.</li><li>Limit: another layer to trust. It decides what to cut, and since the original is always one retrieval away, an agent that keeps asking for it shrinks the savings back down.</li></ul><h3>CodeGraph</h3><ul><li>Best when repo navigation is the problem.</li><li>It builds a local map of your code: symbols, what calls what, and how files depend on each other. So instead of grepping around and opening files one by one, the agent asks once and gets the right code back.</li><li>Its benchmark reports 58% fewer tool calls, 22% faster, and file reads cut to almost zero across 7 repos.</li><li>Limit: the savings depend on repo size and usage volume. On small repos, the difference may be barely noticeable. On large codebases with frequent agent runs, it becomes much more interesting.</li></ul><h3>Ponytail</h3><ul><li>Best when the agent writes too much code.</li><li>It pushes the agent to reuse existing code, platform features, and dependencies before writing anything new.</li><li>In their benchmark: 54% less code, 22% fewer tokens, 20% lower cost, 27% faster, and no safety checks dropped.</li><li>Limit: the savings are uneven. Big when the agent would have over-built, almost nothing when the code is already lean.</li></ul><h3>Caveman</h3><ul><li>Best when the agent writes too much English.</li><li>It cuts verbosity: less filler, shorter answers, more direct responses.</li><li>Reported reduction: around 65% of output tokens.</li><li>Limit: output is usually the cheap part, and reasoning tokens are untouched. In one coding benchmark, using Caveman actually raised total tokens by 7%.</li></ul><h3>My current take:</h3><ul><li>Use RTK for terminal noise.</li><li>Use Headroom for messy multi-source context.</li><li>Use CodeGraph for repo navigation.</li><li>Use Ponytail for smaller diffs.</li><li>Use Caveman for shorter replies.</li></ul><h3>BUT</h3><p>Don’t assume 80% fewer tokens means 80% cheaper.</p><p>Token reduction still matters, but it does not map 1:1 to cost reduction. Cached input, reasoning tokens, and reruns all affect the final bill.</p><p>So the useful question is not: which tool saves the most tokens?</p><p>It is: <strong>where is my agent actually wasting context?</strong></p><p>Measure input tokens, output tokens, reasoning tokens, tool calls, latency, cost per task, tests passed, reruns, and how often the agent loses context.</p><p>The goal is not fewer tokens as a vanity metric.</p><p>The goal is less noise in, better context selected, smaller diffs, fewer reruns, and lower cost per successful task.</p><h3>Sources:</h3><ol><li>RTK - <a href="https://github.com/rtk-ai/rtk">https://github.com/rtk-ai/rtk</a></li><li>Headroom - <a href="https://github.com/chopratejas/headroom">https://github.com/chopratejas/headroom</a></li><li>CodeGraph - <a href="https://github.com/colbymchenry/codegraph">https://github.com/colbymchenry/codegraph</a></li><li>Ponytail - <a href="https://github.com/DietrichGebert/ponytail">https://github.com/DietrichGebert/ponytail</a></li><li>Caveman - <a href="https://github.com/JuliusBrussee/caveman">https://github.com/JuliusBrussee/caveman</a></li></ol><p>The numbers above are from project benchmarks and specific tests, not a guarantee for every setup. Use them as direction, then measure where your own agent actually spends and wastes tokens.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d36eb5b46442" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How I Built a Fully Offline AI App]]></title>
            <link>https://gomzkov.medium.com/how-i-built-a-fully-offline-ai-app-for-private-journaling-25911b448671?source=rss-28ee7e3e3626------2</link>
            <guid isPermaLink="false">https://medium.com/p/25911b448671</guid>
            <category><![CDATA[gemma]]></category>
            <category><![CDATA[langchain]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[chromadb]]></category>
            <category><![CDATA[ollama]]></category>
            <dc:creator><![CDATA[Evgeni Gomziakov]]></dc:creator>
            <pubDate>Tue, 27 May 2025 08:05:49 GMT</pubDate>
            <atom:updated>2025-08-24T15:37:20.049Z</atom:updated>
            <content:encoded><![CDATA[<h4>Building an offline app with <strong>LangChain, </strong>Ollama and Gemma 3 for personalized insights, while keeping all data on your machine.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qfva-KepzRDoNr9CKAE6wA.jpeg" /></figure><p>Code — <a href="https://github.com/zappodev/jbud">https://github.com/zappodev/jbud</a></p><h3>Tech Stack</h3><ul><li><strong>Ollama + Gemma 3</strong> for local LLM processing. No API keys needed.</li><li><strong>ChromaDB</strong> for semantic search through journal entries.</li><li><strong>LangChain</strong> for RAG pipeline — reduced 100+ lines to 20.</li><li><strong>Streamlit</strong> for the UI. Fast prototyping without complexity.</li><li><strong>JSON files</strong> for data storage. Users can see exactly what’s saved.</li></ul><h3>Key Architecture Decision</h3><p>Instead of generic AI responses, I built a RAG system where your journal entries become the knowledge base. After months of entries, ask “What patterns do I see in my productivity?” or “Where do I tend to struggle?” The AI searches your actual experiences and identifies recurring themes, blind spots, and improvement opportunities.</p><p>This approach gives you insights based on your real behavior patterns rather than generic advice.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YRUczhcU9AD9qYXyYBSTdA.png" /></figure><h3>What I Learned</h3><p><strong>LangChain is worth it for RAG.</strong> I considered manual implementation but LangChain’s abstractions saved significant time and complexity.</p><p><strong>Semantic search transforms small apps.</strong> ChromaDB finds “productivity” entries when you wrote “got things done” or “focused well.” Simple keyword search would miss these connections.</p><p><strong>Local LLMs are production-ready.</strong> Ollama setup was smooth. Gemma 3 performs well for personal use cases.</p><p><strong>Privacy-by-design builds trust.</strong> Transparent JSON storage where users control their data. No hidden processing or external calls.</p><h3>The Local-First Advantage</h3><ul><li>No internet required after setup</li><li>Data stays in local folders you control</li><li>No rate limits or API costs</li><li>Personal insights without data exposure</li></ul><h3>What I’d Change</h3><p>Start with simpler MVP using just Ollama and basic prompts. Add ChromaDB later. Include more structured data extraction from entries. Test other local models beyond Gemma 3.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WFCJZFnY6xy7VQT2HmZ07w.png" /></figure><h3>The Bigger Picture</h3><p>Local-first AI is ready for personal applications. The technology is there, setup is simple, and people want AI insights about their daily life without sharing personal data with third parties.</p><p>Most personal AI apps still use client-server architecture. Running LLMs locally opens up applications that were impossible before.</p><p>Stack: Ollama, Gemma 3, LangChain, ChromaDB, Streamlit.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=25911b448671" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Building a CLI App with Node.js]]></title>
            <link>https://gomzkov.medium.com/building-a-cli-with-node-js-in-2024-c278802a3ef5?source=rss-28ee7e3e3626------2</link>
            <guid isPermaLink="false">https://medium.com/p/c278802a3ef5</guid>
            <category><![CDATA[javascript]]></category>
            <category><![CDATA[nodejs]]></category>
            <category><![CDATA[command-line]]></category>
            <category><![CDATA[cli]]></category>
            <category><![CDATA[backend]]></category>
            <dc:creator><![CDATA[Evgeni Gomziakov]]></dc:creator>
            <pubDate>Fri, 09 Feb 2024 08:29:57 GMT</pubDate>
            <atom:updated>2026-02-03T20:25:35.904Z</atom:updated>
            <content:encoded><![CDATA[<h4>An in-depth step-by-step guide to creating a CLI App with Node.js, covering everything from command handling and user prompts to enhancing user experience, as well as organizing project structure and documentation.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CWyxQW1NBDMPtaDE6JwbKg.png" /></figure><h4>Why Node?</h4><p>The event-driven architecture, along with the npm ecosystem that offers many packages designed specifically for this purpose, makes it the go-to choice for developing efficient and scalable CLI tools.</p><h4>Why Build a CLI App?</h4><ol><li>Automate tasks</li><li>Create tools for developers</li><li>Interact with systems and manage flows</li></ol><h4>Real World Examples</h4><p>At Nielsen we created several CLIs that provided huge value:</p><ul><li>A CLI that manages dynamic pipelines in the CI/CD flows — no more manual configuration or waiting between processes<strong>.</strong></li><li>A CLI that sets up and manages local dockerized development environments<strong>.</strong></li></ul><p>Now you’ll see how easy it is to make one.<br>For those eager to dive straight into the code, the files can be found <a href="https://github.com/gomzkov/my-node-cli">here</a>.</p><h3><strong>Table Of Contents</strong></h3><ul><li><a href="#f5aa">Setting Up</a><br> ∘ <a href="#0762">Bootstrap Your Project</a><br> ∘ <a href="#285d">Bring in Commander.js</a><br> ∘ <a href="#de78">Crafting The CLI</a></li><li><a href="#efff">User Experience</a><br> ∘ <a href="#a1f0">Add Some Color</a><br> ∘ <a href="#9df0">Prompting Made Easy</a><br> ∘ <a href="#0d0a">Cool Loaders</a><br> ∘ <a href="#3a95">Adding ASCII Art</a></li><li><a href="#9781">Project Structure</a></li><li><a href="#35f3">Documentation</a><br> ∘ <a href="#dbe1">Auto Generating Documentation</a></li><li><a href="#b1da">Best Practices</a></li></ul><h3>Setting Up</h3><p>make sure you’ve got <a href="https://nodejs.org/en"><strong>Node.js</strong></a><strong> </strong>on your machine.</p><h4>Step 1: Bootstrap Your Project</h4><p>Create a new folder for your project and jump into it:</p><pre>mkdir my-node-cli<br>cd my-node-cli</pre><p>Fire up a new Node.js project:</p><pre>npm init</pre><h4>Step 2: Bring in Commander.js</h4><p><a href="https://github.com/tj/commander.js">Commander.js</a> is our go-to for building CLIs in Node.js. It’s like having a swiss army knife for parsing input, help text, and managing errors.</p><pre>npm install commander</pre><h4>Step 3: Crafting The CLI</h4><p>Create a file named <strong>index.js</strong> in your project folder. This will be where our CLI starts. Add a shebang at the top to get this CLI off the ground.</p><pre>#!/usr/bin/env node<br><br>import { program } from &quot;commander&quot;;<br><br>program<br>  .version(&quot;1.0.0&quot;)<br>  .description(&quot;My Node CLI&quot;)<br>  .option(&quot;-n, --name &lt;type&gt;&quot;, &quot;Add your name&quot;)<br>  .action((options) =&gt; {<br>    console.log(`Hey, ${options.name}!`);<br>  });<br><br>program.parse(process.argv);</pre><p>Add <strong>bin</strong> to your package.json to recognize your CLI command, and <strong>type</strong> to work with ES modules instead of CommonJS:</p><pre>&quot;bin&quot;: {<br>    &quot;my-node-cli&quot;: &quot;./index.js&quot;<br>},<br>&quot;type&quot;: &quot;module&quot;</pre><p>Link your project globally with:</p><pre>npm link </pre><p>And just like that, <strong>my-node-cli</strong> is ready to run on your terminal!</p><pre>my-node-cli --name YourName</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0fgpq_666Q7lS1i_vt_M1g.png" /></figure><p><strong>Note</strong>: since Node.js 18.3 we have <strong>built in command-line arguments parser</strong>. You can read about it <a href="https://nodejs.org/api/util.html#utilparseargsconfig">here</a> and decide if you want to use it instead of commander.js.</p><h3>User Experience</h3><h4>Add Some Color</h4><p><a href="https://github.com/chalk/chalk">Chalk</a> is perfect for making your CLI’s output pop with color. Grab it with:</p><pre>npm install chalk</pre><p>Now, let’s improve our greeting:</p><pre>#!/usr/bin/env node<br><br>import { program } from &quot;commander&quot;;<br>import chalk from &quot;chalk&quot;;<br><br>program<br>  .version(&quot;1.0.0&quot;)<br>  .description(&quot;My Node CLI&quot;)<br>  .option(&quot;-n, --name &lt;type&gt;&quot;, &quot;Add your name&quot;)<br>  .action((options) =&gt; {<br>    console.log(chalk.blue(`Hey, ${options.name}!`));<br>    console.log(chalk.green(`Hey, ${options.name}!`));<br>    console.log(chalk.red(`Hey, ${options.name}!`));<br>  });<br><br>program.parse(process.argv);</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*w7lU61FIDimij8HKLlTVbg.png" /></figure><h4>Prompting Made Easy</h4><p>For a more interactive vibe, <a href="https://github.com/SBoudrias/Inquirer.js"><strong>Inquirer.js</strong></a> is your friend.</p><pre>npm install inquirer</pre><p>Instead of using command line options to collect data — just ask the user</p><pre>#!/usr/bin/env node<br><br>import { program } from &quot;commander&quot;;<br>import chalk from &quot;chalk&quot;;<br>import inquirer from &quot;inquirer&quot;;<br><br>program.version(&quot;1.0.0&quot;).description(&quot;My Node CLI&quot;);<br><br>program.action(() =&gt; {<br>  inquirer<br>    .prompt([<br>      {<br>        type: &quot;input&quot;,<br>        name: &quot;name&quot;,<br>        message: &quot;What&#39;s your name?&quot;,<br>      },<br>    ])<br>    .then((answers) =&gt; {<br>      console.log(chalk.green(`Hey there, ${answers.name}!`));<br>    });<br>});<br><br>program.parse(process.argv);</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2w1hpzni9zecIeYR6DY7Ww.png" /></figure><p>There is a <strong>Confirm</strong> prompt type<strong>— </strong>Asks the user a yes/no question.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*geB0FNxg6Derju_kTIyDfA.png" /></figure><p><strong>List </strong>prompt type — Allows the user to choose from a list of options.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OXAH_xJeaq1ev5CSrAnFRQ.png" /></figure><p>And there are also <strong>Checkbox, Password</strong>, <strong>Rawlist </strong>and <strong>Expand</strong>. Feel free to explore more at <a href="https://github.com/SBoudrias/Inquirer.js">https://github.com/SBoudrias/Inquirer.js</a></p><h4>Cool Loaders</h4><p>Loading times? Make them fun with <a href="https://github.com/sindresorhus/ora"><strong>ora</strong></a>. It’s great for adding spinner animations:</p><pre>npm install ora</pre><p>Sprinkle in a loader for processes that take time:</p><pre>#!/usr/bin/env node<br><br>import { program } from &quot;commander&quot;;<br>import chalk from &quot;chalk&quot;;<br>import inquirer from &quot;inquirer&quot;;<br>import ora from &quot;ora&quot;;<br><br>program.version(&quot;1.0.0&quot;).description(&quot;My Node CLI&quot;);<br><br>program.action(() =&gt; {<br>  inquirer<br>    .prompt([<br>      {<br>        type: &quot;list&quot;,<br>        name: &quot;choice&quot;,<br>        message: &quot;Choose an option:&quot;,<br>        choices: [&quot;Option 1&quot;, &quot;Option 2&quot;, &quot;Option 3&quot;],<br>      },<br>    ])<br>    .then((result) =&gt; {<br>      const spinner = ora(`Doing ${result.choice}...`).start(); // Start the spinner<br><br>      setTimeout(() =&gt; {<br>        spinner.succeed(chalk.green(&quot;Done!&quot;));<br>      }, 3000);<br>    });<br>});<br><br>program.parse(process.argv);</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*wFmspYHAW5572fRutff1MQ.gif" /></figure><h4>Adding ASCII Art</h4><p>Let’s add some final touches with <a href="https://github.com/patorjk/figlet.js">figlet.js</a>:</p><pre>npm install figlet</pre><p>Add this to your <strong>index.js</strong></p><pre>import figlet from &quot;figlet&quot;;<br><br>console.log(<br>  chalk.yellow(figlet.textSync(&quot;My Node CLI&quot;, { horizontalLayout: &quot;full&quot; }))<br>);</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*omuOfRQmUtSo7EbmJKuJXA.png" /></figure><p>There are various fonts and customization options, allowing you to tailor the ASCII art to your CLI’ aesthetic.</p><h3>Project Structure</h3><p>Keeping things organized can save you a ton of time later on, especially as your project grows. Here’s a simple yet effective structure to start with:</p><pre>my-node-cli/<br>├─ bin/<br>│ └─ index.js<br>├─ src/<br>│ ├─ commands/<br>│ ├─ utils/<br>│ └─ lib/<br>├─ package.json<br>└─ README.md</pre><ul><li><strong>bin — </strong>is where your CLI’s lives. It’s what gets called when someone runs your CLI.</li><li><strong>src/commands — </strong>holds individual command files. This makes adding new commands or editing existing ones cleaner.</li><li><strong>src/utils — </strong>is for utility functions you might need across several commands, like formatting data.</li><li><strong>src/lib — </strong>could be where your core functionality resides, especially if your CLI interacts with APIs or performs complex logic.</li></ul><h3>Documentation</h3><p>Clear documentation is key. Outline installation, usage, and command options in your README.md to guide users through your CLI tool.</p><pre># My Node CLI<br>My Node CLI is a tool for doing awesome things directly from your terminal.<br><br>## Installation<br><br>```bash<br>npm install -g my-node-cli<br>```<br><br>## Usage<br>To start using My Node CLI, run:<br><br>```bash<br>my-node-cli - help<br>```<br><br>### Commands<br>- `my-node-cli - name YourName`: Greets you by your name.<br>- `my-node-cli option1`: Executes option 1.<br><br>For more detailed information on commands, run `my-node-cli --help`.<br><br>## Contributing<br>Contributions are welcome ...<br><br>## License<br>This project is licensed ...</pre><h4>Auto Generating Documentation</h4><p>Consider using tools like <a href="https://github.com/jsdoc/jsdoc"><strong>JSDoc</strong></a> or <a href="https://github.com/TypeStrong/typedoc"><strong>TypeDoc</strong></a>, they can generate detailed documentation from your code comments.</p><pre>/**<br> * This function greets the user by name.<br> * @param {string} name The name of the user.<br> */<br>const greet = (name) =&gt; {<br>  console.log(`Hello, ${name}!`);<br>};</pre><h3>Best Practices</h3><p>Before you start working on the actual CLI logic I would strongly recommend checking this <a href="https://github.com/lirantal/nodejs-cli-apps-best-practices">repo by Liran Tal</a>, it has more than 3k stars and covers all the best practices I could think of and more.</p><p>For example, instead of requiring your user to repeatedly provide the same information between invocation, provide a stateful experience using <a href="https://github.com/sindresorhus/conf">conf</a>, for saving data like username, email or API tokens.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zDZ5lKnlj-PIqtk1r8qPSw.png" /></figure><p>Ready to see all this in action? Check out the complete project along with all the example files on my <a href="https://github.com/gomzkov/my-node-cli">GitHub page</a>. Dive in, play around, and don’t hesitate to fork or star the repo if you find it helpful!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c278802a3ef5" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Practical Habits for Better Coding]]></title>
            <link>https://levelup.gitconnected.com/4-habits-that-made-me-a-better-developer-39b647903534?source=rss-28ee7e3e3626------2</link>
            <guid isPermaLink="false">https://medium.com/p/39b647903534</guid>
            <category><![CDATA[tools-for-developers]]></category>
            <category><![CDATA[developer]]></category>
            <category><![CDATA[productivity]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[programming]]></category>
            <dc:creator><![CDATA[Evgeni Gomziakov]]></dc:creator>
            <pubDate>Mon, 26 Sep 2022 21:29:42 GMT</pubDate>
            <atom:updated>2024-11-24T08:54:29.593Z</atom:updated>
            <content:encoded><![CDATA[<h4>Simple tips that improved my coding experience</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1013/1*8ESkMhx0ftmLm1NpDYc2wA.png" /></figure><h3>1. Keep it Simple</h3><p>In my coding journey, I’ve found that simplicity often leads to better results. If I’m spending too much time trying to figure out a piece of code, it’s usually a sign that I need to simplify. Here’s how I approach it:</p><ul><li><strong>Break Down Problems</strong>: try to divide complex problems into smaller, more manageable pieces.</li><li><strong>Avoid Over-Engineering</strong>: focus on what’s needed now rather than adding complexity for future “what ifs.”</li><li><strong>Refactor Regularly</strong>: keep an eye out for opportunities to simplify my code, even after it’s working.</li></ul><p>I’ve often started with something overly complex for a basic task, like this:</p><pre>type Numeric = number | string;<br><br>interface AddOptions {<br>  checkTypes?: boolean;<br>  round?: boolean;<br>  precision?: number;<br>}<br><br>function addComplex&lt;T extends Numeric, U extends Numeric&gt;(a: T, b: U, options: AddOptions = {}): number | string {<br>  if (options.checkTypes) {<br>    const isANumeric = typeof a === &#39;number&#39; || (typeof a === &#39;string&#39; &amp;&amp; !isNaN(Number(a)));<br>    const isBNumeric = typeof b === &#39;number&#39; || (typeof b === &#39;string&#39; &amp;&amp; !isNaN(Number(b)));<br>    if (!isANumeric || !isBNumeric) {<br>      throw new Error(&#39;Invalid type: only numeric values or numeric strings are allowed&#39;);<br>    }<br>  }<br><br>  let result = Number(a) + Number(b);<br><br>  if (options.round) {<br>    result = Math.round(result);<br>  } else if (typeof options.precision === &#39;number&#39;) {<br>    result = parseFloat(result.toFixed(options.precision));<br>  }<br><br>  return result;<br>}</pre><p>But often, a simple version is all that’s needed:</p><pre>const addSimple = (a,b) =&gt; (a + b)</pre><p>Starting simple and only adding complexity when absolutely necessary has saved me a lot of headaches.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*DFWNzQbr15AVgHqF2LXAlg.png" /></figure><h3>2. Keep Up with Tech Stuff</h3><p>One habit that’s really helped me grow is setting aside a little time each week to learn something new. Whether it’s a new framework, a coding challenge, or just following a tech blog, it all adds up over time.</p><p><strong>Here are a few ways I try to stay in the loop:</strong></p><ul><li>Try Coding Challenges</li><li>Follow Tech Blogs and Podcasts</li><li>Join Developer Communities</li><li>Experiment with New Tools</li></ul><p><strong>If you’re looking for places to start, here are some of my go-tos:</strong></p><ol><li><a href="https://www.youtube.com/@Fireship"><strong>Fireship</strong></a><strong> </strong>— quick tutorials and insights into the latest technologies</li><li><a href="https://dev.to/"><strong>Dev.to</strong></a> — community where developers share knowledge</li><li><a href="https://github.com/trending?since=weekly"><strong>Trending GitHub Repositories</strong></a> — great way to keep an eye on what’s new</li><li><a href="https://daily.dev/"><strong>Daily.dev</strong></a> — extension that turns new tabs into a collection of tech news</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*bR_dM8VqNFwQhJ9N2vf2Sw.png" /></figure><h3>3. Document Your Learnings</h3><p>One of the most effective ways I’ve found to solidify what I learn is by documenting it. Writing things down not only helps me understand concepts better but also creates a<strong> resource that I can refer back to and share with others</strong>.</p><p><strong>Keep a Dev Journal — </strong>I try to sum up key takeaways from what I learn each day in <a href="https://www.notion.so/"><strong>Notion</strong></a>. It could be a new concept, a tricky bug I solved, or just reflections on my progress.</p><p><strong>Create Tutorials or Guides — </strong>Sometimes, I find that the best way to fully grasp a concept is by teaching it to others. Creating tutorials forces me to break down complex ideas into simple, understandable steps.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*I2XaTPwwLP7J34bF" /></figure><h3>4. <strong>Focus on Code Quality</strong></h3><p>It’s easy to rush through coding tasks, but I’ve learned that taking the time to focus on code quality can save a lot of trouble down the line.</p><p><strong>Adopt Best Practices</strong> — I regularly check whether my code adheres to established best practices, such as <strong>Clean Code</strong>, <strong>SOLID</strong> principles, <strong>DRY</strong> (Don’t Repeat Yourself), and proper error handling.</p><p><strong>Code Reviews</strong> — I learn a lot by reviewing others’ code and getting feedback on mine. It’s a two-way street that benefits everyone.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*810j7sg4t04mnO1G.png" /></figure><p><strong>Thanks for reading</strong>! I hope you found these tips useful. Feel free to share your thoughts or pass this article along to others who might find it helpful. <strong>Happy coding</strong>!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=39b647903534" width="1" height="1" alt=""><hr><p><a href="https://levelup.gitconnected.com/4-habits-that-made-me-a-better-developer-39b647903534">Practical Habits for Better Coding</a> was originally published in <a href="https://levelup.gitconnected.com/">Level Up Coding</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>