<?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 Grotti on Medium]]></title>
        <description><![CDATA[Stories by Grotti on Medium]]></description>
        <link>https://medium.com/@grotti.?source=rss-ff39cf4a8964------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*WRsOejLjy2oKfMiyscn7lg.png</url>
            <title>Stories by Grotti on Medium</title>
            <link>https://medium.com/@grotti.?source=rss-ff39cf4a8964------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 05:04:11 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@grotti./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[I was a speaker at Splunk .conf25, so what?]]></title>
            <link>https://medium.com/@grotti./i-was-a-speaker-at-splunk-conf25-so-what-1506662c5fe1?source=rss-ff39cf4a8964------2</link>
            <guid isPermaLink="false">https://medium.com/p/1506662c5fe1</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[splunk]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[conference]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Grotti]]></dc:creator>
            <pubDate>Sat, 29 Nov 2025 14:35:32 GMT</pubDate>
            <atom:updated>2025-11-29T14:35:32.408Z</atom:updated>
            <content:encoded><![CDATA[<h3>I was a speaker at Splunk .conf25, so what?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*R6eX7E1GBHXnF4fILBYgIQ.jpeg" /><figcaption>Image of Boston city.</figcaption></figure><p>Most people go to <strong>Splunk .conf</strong> for the first time to watch, learn, and soak in the atmosphere.</p><p>I didn’t.</p><p>My first time at .conf was already on stage as a <strong>speaker</strong>. No warm-up year, no “maybe next time,” no easing into the experience. Just straight to the spotlight, the mic, and a room full of people expecting to learn something meaningful.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BE-EAhxD-g6qPI8fUOnqFA.jpeg" /></figure><p>And honestly? That made Splunk .conf25 feel even more surreal.</p><p>Walking into the event for the very first time felt like stepping into a massive playground built for people who breathe data: gigantic screens, neon-lit corridors, hands-on labs everywhere, engineers from every corner of the world, and that unmistakable fusion of Splunk humor, chaos, and deep technical obsession. If someone had told me years ago that my first time at .conf wouldn’t be as an attendee, but already speaking maybe I would have laughed.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*E9NqUjdROmfwbRjddQnYpA.jpeg" /></figure><p>Yet there I was.</p><p>And while I was waiting to go on stage, one thought kept echoing:<br> <strong>“Okay… it’s my first .conf ever, and I’m already a speaker. So what?”</strong></p><p>Because the real meaning wasn’t the badge or the title. It was contributing back to a community that taught me so much <strong>(and between us, the Splunk community is huge and very strong)</strong> , sharing lessons that came from real challenges, real architectures, real life experience, the kind of knowledge you only get after living inside a SOC long enough.</p><p>.conf25 wasn’t just an event for me. It was a milestone, a mix of adrenaline, gratitude, and the realization that sometimes your first step into something big can already be a giant leap.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JCxWYEp_pBYcQZA0bPfePw.jpeg" /><figcaption>Me and Victor at our Splunk .conf Session</figcaption></figure><p>This was my first time at .conf, it was also my first time experiencing <strong>Boston, </strong>and the city ended up becoming part of the story.</p><p>Boston isn’t just the place where .conf25 happened. It sets the tone. The moment you step outside the convention center, you feel that mix of history and innovation everywhere: old brick buildings sitting right next to labs, startups, and some of the smartest institutions in the world. It’s a city that feels academic, but alive, traditional and historical.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bDgHyEZG0gTJPmDtrGrb-Q.jpeg" /></figure><p>And as someone coming in for their first .conf (and first time as a speaker), Boston almost felt symbolic. There’s something inspiring about giving a talk in a place surrounded by MIT, Harvard, and a whole ecosystem.</p><p>But it wasn’t just the intellectual vibe. Boston is <em>beautiful</em>. From the waterfront views in the Seaport District, to the quiet morning walks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*h1X8pu8iLgpCUpEZ.jpg" /><figcaption>Samuel Adams Beer is GREAT</figcaption></figure><p>Particularly, I loved Boston Downtown!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*PymY8ptiioQQiOg-.jpg" /></figure><p>There was something almost cinematic about leaving the conference after a long day of sessions and stepping into those clean, cold streets illuminated by the reflection of the harbor. For someone experiencing .conf for the first time, the city made the whole thing feel even bigger, even more meaningful.</p><p>The Lucas of 2016 could never have imagined that he would experience something like this one day. It’s in moments like these that I remember my first encounters with a computer and how I fell in love with it and could never see myself doing anything else related to it</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Bu8jhlty_xvFKTuFzI7AEg.jpeg" /></figure><p>My session was about SOAR and automation for SOC. If you’d like to learn more, the link to the material is on the Splunk website, here → <a href="https://conf.splunk.com/files/2025/slides/SEC1364.pdf">https://conf.splunk.com/files/2025/slides/SEC1364.pdf</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/766/1*ViKSkFeBJOX55AnaQRQUJA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VA77hlHn8vPDKYY-FRGdrw.jpeg" /><figcaption>At .Conf with colleagues from C6 Bank, NTT Data, PagSeguro, Splunk, TD Sinnex and DataRunk</figcaption></figure><p>After 2 months trying to assimilate this whole experience, I think i can describe with words all I faced and experienced, just random toughs about a brazilian young guy who speaked in Splunk .conf 2025 :) 🙃</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1506662c5fe1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Exploring the Splunk AI Toolkit (f.k.a]]></title>
            <link>https://medium.com/@grotti./exploring-the-splunk-machine-learning-toolkit-part-1-outlier-detection-in-application-requests-5e1e047f6b99?source=rss-ff39cf4a8964------2</link>
            <guid isPermaLink="false">https://medium.com/p/5e1e047f6b99</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[splunk]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[coding]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Grotti]]></dc:creator>
            <pubDate>Mon, 29 Sep 2025 00:31:45 GMT</pubDate>
            <atom:updated>2025-09-29T11:54:06.664Z</atom:updated>
            <content:encoded><![CDATA[<h3>Exploring the Splunk AI Toolkit (f.k.a Splunk Machine Learning Toolkit): Outlier Detection in Application Requests</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/950/0*Pi9lrDhIHlEMsB9H.png" /></figure><p>Splunk is well known for its powerful search capabilities and data analytics features. But what many practitioners don’t realize is that Splunk can also be extended with <strong>machine learning</strong>, using <strong>“Splunk Machine Learning Toolkit”</strong> to detect patterns, anomalies, and even predict future behavior. In Brazil, i don’t see this type of features very much in use, primarily by SOCs and detection engineering teams.</p><p>The Splunk Machine Learning Toolkit (MLTK) is an app available for both Splunk Enterprise and Splunk Cloud Platform users through <a href="https://splunkbase.splunk.com/app/2890">Splunkbase</a>. The Splunk Machine Learning Toolkit acts like an extension to the Splunk platform and includes machine learning SPL search commands, macros, and visualizations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/950/0*RhHvWSJ-Oi9YAGvL.png" /></figure><p>The first step is choosing what model you want to use, in this case, i will show how to use the “Smart Outlier Detection” model, to detect numeric outliers in application requests.</p><p>It’s easy to implement, using only a little of SPL.</p><h3>Why Numerical Outlier Detection Matters?</h3><p>When working with applications, the majority of requests usually follow predictable patterns, request counts fluctuate around an average, and error rates remain stable.</p><p>But sometimes, requests spike abnormally, errors skyrocket, or response times become unusually high. These <strong>outliers</strong> may indicate:</p><ul><li><strong>Security incidents</strong> (brute-force attempts, DoS attacks, or API abuse and enumeration)</li><li><strong>Performance problems</strong> (bottlenecks or degraded services)</li><li><strong>Unexpected user behavior</strong> (misuse or misconfiguration)</li></ul><p>Identifying these anomalies early allows teams to react quickly, minimizing downtime and improving resilience.</p><h3>Data Preparation</h3><p>First, you need to extract the data, right? I recommend you to grab some WAF logs or another application logs, but, is important to have a quantity of the requests in those logs and the path which those requests are made, ok?</p><p><strong>SPL</strong> like this is good to go:</p><pre>index=waf_logs http_host=&quot;ecommerce.com&quot; application_path=TERM(/store/login/) OR application_path=TERM(/store/signup/)<br>| fields application_path , _time<br>| bin _time span=15m<br>| stats count by _time, application_path</pre><p>It is highly recommended that you point out the routes that you want the model to detect numerical outliers in the requests, limit the search by the scope of routes, it is possible to automate this, but this will be for another topic, in this SPL example above I am using “login” and “registration” pseudo routes because a very common behavior is attackers making a high number of requests on these routes.</p><p>After running this SPL you will have an idea of what the final output will look like. Another tip is to put the results in a lookup, as we will use this lookup to feed the model.</p><p>Using an SPL like this you put it in a lookup, already processed and ready to feed the model:</p><pre>index=waf_logs http_host=&quot;ecommerce.com&quot; application_path=TERM(/store/login/) OR application_path=TERM(/store/signup/)<br>| fields application_path, _time<br>| bin _time span=15m<br>| stats count by _time, application_path<br>| outputlookup outlier_detection_ecommerce.csv</pre><p>Your lookup will look something like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*p9C4XjnkzSzICeG1FYDt4Q.png" /><figcaption><a href="https://splunkbase.splunk.com/app/1724">Splunk App for Lookup File Editing | Splunkbase</a></figcaption></figure><p>basically request count grouped by application route and a 15 minutes time frame.</p><h3><strong>Feeding the model</strong></h3><p>Now, finally, let’s go to the Splunk Machine Learning Toolkit app.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VHEak9Qouf1Ad6HIbO4OVg.png" /></figure><p>Once you open the <strong>Machine Learning Toolkit app</strong>, you’ll find the <strong>Showcase</strong> a collection of interactive examples that demonstrate how different machine learning operations can be applied directly to your Splunk data. These examples are not just “toy models”; they’re pre-built, end-to-end workflows that allow you to inspect the SPL, the algorithms, and even the Python code behind them.</p><p>Here’s a quick overview of what you’ll see in the Showcase:</p><h4>LLM Integrations</h4><p>Splunk now allows you to enrich your data using <strong>externally hosted Large Language Models (LLMs)</strong>.</p><h4>Predict Fields</h4><p>This section focuses on <strong>predictive modeling</strong>. You can train models to forecast the value of a numeric or categorical field based on other attributes in your data.<br> Examples include:</p><h4>Detect Outliers (We are going to use this!)</h4><p>Outlier detection is one of the most practical use cases in security and IT operations. With <strong>14 examples</strong>, the Showcase demonstrates how to find data points that deviate significantly from normal behavior.<br> This can help you uncover:</p><h4>Forecast Time Series</h4><p>With <strong>9 examples available</strong>, this category demonstrates how to predict future values of time-series data using past observations.<br> Common use cases:</p><h4>Cluster Events</h4><p>The <strong>Clustering</strong> section shows how Splunk can automatically group similar events together.</p><p>Under the hood, these examples use algorithms like <strong>k-means, hierarchical clustering, and density-based clustering</strong>.</p><p>Ok, lets continue…</p><p>First of all, you need to go to the “Experiments” tab and create a new experiment using the <strong>Smart Outlier Detection Model.</strong></p><p>Too feed the model with your dataset, use this example of SPL:</p><pre>| inputlookup outlier_detection_ecommerce.csv</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QXm6a5yC4FXaRMpe7MFuGA.png" /><figcaption>Feeding the model</figcaption></figure><p>Click in the “Next” button to go to the analyze e processing phase.</p><p>The field that we want to identify numerical outliers is the <em>count</em> field.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*laLpPBbu2xS0gnZt8vCaPw.png" /></figure><p>And, when you select the field you want to analyze and click on “Detect Outliers”, the process of detecting outliers for itself starts. After that, you can see in the image above that <strong>8 Outliers were detected in our dataset.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qmYYY8TXpWTruoRZgqCBCQ.png" /></figure><p>And arriving at the last step, you can operationalize this detection, you can publish the experiment, generate an alert from the experiment, to generate in your operation for example.</p><p><strong>Very easy that even your grandma can use the MLTK models :)</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*W9Gt2HcopRRbK01u8mbmFw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/598/1*asdgDG9KsBX6YuuyphV8Nw.png" /></figure><p>And finally, I wanted to show how you can leverage the Machine Learning models available in the <strong>Splunk Machine Learning Toolkit (MLTK)</strong>. This app makes it easy and quick to apply ML in your environment. Surprisingly, the Splunk Machine Learning Toolkit isn’t heavily promoted by Splunk itself, and I’ve noticed that many people, especially here in Brazil, don’t really use it.</p><p>The main idea behind writing these kinds of articles is to share knowledge and help others take full advantage of the tools they already have. The Splunk Machine Learning Toolkit is one of them: it’s completely free on Splunkbase (<a href="https://splunkbase.splunk.com/app/2890">Splunk AI Toolkit | Splunkbase</a>) , and you can install it on both <strong>Splunk Enterprise</strong> and <strong>Splunk Cloud</strong>. From there, you can operationalize ML by creating alerts and detections with the prebuilt models included in the app.</p><p>This can power operations across different areas, infrastructure ops, observability, SOC, or security ops in general.</p><p>That’s it for now! Thank you so much for reading. I’ll be publishing more articles like this one this is just the first of many.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5e1e047f6b99" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>