<?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 Amina Okanovic on Medium]]></title>
        <description><![CDATA[Stories by Amina Okanovic on Medium]]></description>
        <link>https://medium.com/@kabirbakovic?source=rss-84c59f25e34d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*Q2NwUZQep2q9muKXURS5Yw.jpeg</url>
            <title>Stories by Amina Okanovic on Medium</title>
            <link>https://medium.com/@kabirbakovic?source=rss-84c59f25e34d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 03:15:31 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@kabirbakovic/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[The Expensive Part of Entity Resolution Is the Step Nobody Budgets Time For]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@kabirbakovic/the-expensive-part-of-entity-resolution-is-the-step-nobody-budgets-time-for-28ccdb42dd29?source=rss-84c59f25e34d------2"><img src="https://cdn-images-1.medium.com/max/2180/1*W4l4F1wbCgODJMSj0qROvA.png" width="2180"></a></p><p class="medium-feed-snippet">An investing knowledge graph pulls company data from a Bloomberg-style feed, a regional-language news service, and a handful of smaller&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@kabirbakovic/the-expensive-part-of-entity-resolution-is-the-step-nobody-budgets-time-for-28ccdb42dd29?source=rss-84c59f25e34d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@kabirbakovic/the-expensive-part-of-entity-resolution-is-the-step-nobody-budgets-time-for-28ccdb42dd29?source=rss-84c59f25e34d------2</link>
            <guid isPermaLink="false">https://medium.com/p/28ccdb42dd29</guid>
            <category><![CDATA[entity-resolution]]></category>
            <category><![CDATA[data-engineering]]></category>
            <category><![CDATA[data-quality]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[data-integration]]></category>
            <dc:creator><![CDATA[Amina Okanovic]]></dc:creator>
            <pubDate>Sat, 03 Oct 2026 07:08:22 GMT</pubDate>
            <atom:updated>2026-10-03T07:08:22.890Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[A Bloated PostgreSQL Table and a Fast One Can Be the Same Table]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@kabirbakovic/a-bloated-postgresql-table-and-a-fast-one-can-be-the-same-table-a7ee141bb70a?source=rss-84c59f25e34d------2"><img src="https://cdn-images-1.medium.com/max/1343/1*CJThVyyX1QDCqgGXIZhWGg.png" width="1343"></a></p><p class="medium-feed-snippet">An engineer checks pg_stat_user_tables on a busy orders table and finds n_dead_tup sitting at several times n_live_tup &#x2014; a table that&apos;s&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@kabirbakovic/a-bloated-postgresql-table-and-a-fast-one-can-be-the-same-table-a7ee141bb70a?source=rss-84c59f25e34d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@kabirbakovic/a-bloated-postgresql-table-and-a-fast-one-can-be-the-same-table-a7ee141bb70a?source=rss-84c59f25e34d------2</link>
            <guid isPermaLink="false">https://medium.com/p/a7ee141bb70a</guid>
            <dc:creator><![CDATA[Amina Okanovic]]></dc:creator>
            <pubDate>Sat, 03 Oct 2026 07:07:18 GMT</pubDate>
            <atom:updated>2026-10-03T07:07:18.540Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[The Real Failure Mode of a Graph RAG System Is Retrieval That Never Stops]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@kabirbakovic/the-real-failure-mode-of-a-graph-rag-system-is-retrieval-that-never-stops-2a2adbb0b6f7?source=rss-84c59f25e34d------2"><img src="https://cdn-images-1.medium.com/max/2229/1*MqyStykHPnf-T9KlTr-9Xw.png" width="2229"></a></p><p class="medium-feed-snippet">A Graph RAG system built over a 50-million-row supply chain knowledge graph starts simple: given a query, find the entities it mentions&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@kabirbakovic/the-real-failure-mode-of-a-graph-rag-system-is-retrieval-that-never-stops-2a2adbb0b6f7?source=rss-84c59f25e34d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@kabirbakovic/the-real-failure-mode-of-a-graph-rag-system-is-retrieval-that-never-stops-2a2adbb0b6f7?source=rss-84c59f25e34d------2</link>
            <guid isPermaLink="false">https://medium.com/p/2a2adbb0b6f7</guid>
            <dc:creator><![CDATA[Amina Okanovic]]></dc:creator>
            <pubDate>Fri, 25 Sep 2026 11:12:14 GMT</pubDate>
            <atom:updated>2026-09-25T11:12:14.525Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Your Data-Quality Tool’s SELECT * Is a BigQuery Cost Bug Wearing a Debugging Feature]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@kabirbakovic/your-data-quality-tools-select-is-a-bigquery-cost-bug-wearing-a-debugging-feature-27a9732a8f6e?source=rss-84c59f25e34d------2"><img src="https://cdn-images-1.medium.com/max/1406/1*M-dTHnMM9qY43-5oduiXxA.png" width="1406"></a></p><p class="medium-feed-snippet">A data-quality check on a 48-column, 5-billion-row BigQuery table fails: one column, amount, has 21 rows with negative values it shouldn&apos;t&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@kabirbakovic/your-data-quality-tools-select-is-a-bigquery-cost-bug-wearing-a-debugging-feature-27a9732a8f6e?source=rss-84c59f25e34d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@kabirbakovic/your-data-quality-tools-select-is-a-bigquery-cost-bug-wearing-a-debugging-feature-27a9732a8f6e?source=rss-84c59f25e34d------2</link>
            <guid isPermaLink="false">https://medium.com/p/27a9732a8f6e</guid>
            <category><![CDATA[sql]]></category>
            <category><![CDATA[bigquery]]></category>
            <category><![CDATA[data-engineering]]></category>
            <category><![CDATA[data-quality]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Amina Okanovic]]></dc:creator>
            <pubDate>Fri, 25 Sep 2026 11:11:42 GMT</pubDate>
            <atom:updated>2026-09-25T11:11:42.720Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[A Data-Drift Detector Passed 56 Fault-Injection Tests. The One Test Expecting Silence Failed.]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@kabirbakovic/a-data-drift-detector-passed-56-fault-injection-tests-the-one-test-expecting-silence-failed-b048932dcd85?source=rss-84c59f25e34d------2"><img src="https://cdn-images-1.medium.com/max/2376/1*F4rR4ewrx_AzkhfUCMeudg.png" width="2376"></a></p><p class="medium-feed-snippet">The team behind Upstrace, an open-source, MIT-licensed column-level drift detector for dbt projects, runs it against a benchmark of 56&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@kabirbakovic/a-data-drift-detector-passed-56-fault-injection-tests-the-one-test-expecting-silence-failed-b048932dcd85?source=rss-84c59f25e34d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@kabirbakovic/a-data-drift-detector-passed-56-fault-injection-tests-the-one-test-expecting-silence-failed-b048932dcd85?source=rss-84c59f25e34d------2</link>
            <guid isPermaLink="false">https://medium.com/p/b048932dcd85</guid>
            <category><![CDATA[duckdb]]></category>
            <category><![CDATA[data-engineering]]></category>
            <category><![CDATA[testing]]></category>
            <category><![CDATA[data-quality]]></category>
            <category><![CDATA[dbt]]></category>
            <dc:creator><![CDATA[Amina Okanovic]]></dc:creator>
            <pubDate>Wed, 23 Sep 2026 06:14:18 GMT</pubDate>
            <atom:updated>2026-09-23T06:14:18.246Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[The Model Change Was the Easy Part: What Netflix Had to Log Before It Could Trust the Result]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@kabirbakovic/the-model-change-was-the-easy-part-what-netflix-had-to-log-before-it-could-trust-the-result-019f671eeb18?source=rss-84c59f25e34d------2"><img src="https://cdn-images-1.medium.com/max/600/1*EMqCpnC0_wf0-gaVDqECNQ.png" width="600"></a></p><p class="medium-feed-snippet">Netflix was about to ship its largest TV home-screen redesign in a decade. The redesign made one artwork canvas &#x2014; short-panel &#x2014; the&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@kabirbakovic/the-model-change-was-the-easy-part-what-netflix-had-to-log-before-it-could-trust-the-result-019f671eeb18?source=rss-84c59f25e34d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@kabirbakovic/the-model-change-was-the-easy-part-what-netflix-had-to-log-before-it-could-trust-the-result-019f671eeb18?source=rss-84c59f25e34d------2</link>
            <guid isPermaLink="false">https://medium.com/p/019f671eeb18</guid>
            <category><![CDATA[data-engineering]]></category>
            <category><![CDATA[vector-embeddings]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[machine-learning-ai]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Amina Okanovic]]></dc:creator>
            <pubDate>Wed, 23 Sep 2026 06:13:45 GMT</pubDate>
            <atom:updated>2026-09-23T06:13:45.144Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[One Connection Per Tenant Does Not Isolate DuckDB, and the Latency Numbers Hide It]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@kabirbakovic/one-connection-per-tenant-does-not-isolate-duckdb-and-the-latency-numbers-hide-it-6136b7344315?source=rss-84c59f25e34d------2"><img src="https://cdn-images-1.medium.com/max/1340/1*2O6kg1nuKrOI2Xf0tR7Bcg.png" width="1340"></a></p><p class="medium-feed-snippet">You have a few hundred gigabytes of Parquet and a DuckDB file that answers questions about it in milliseconds. Six internal teams want SQL&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@kabirbakovic/one-connection-per-tenant-does-not-isolate-duckdb-and-the-latency-numbers-hide-it-6136b7344315?source=rss-84c59f25e34d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@kabirbakovic/one-connection-per-tenant-does-not-isolate-duckdb-and-the-latency-numbers-hide-it-6136b7344315?source=rss-84c59f25e34d------2</link>
            <guid isPermaLink="false">https://medium.com/p/6136b7344315</guid>
            <category><![CDATA[apache-parquet]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[duckdb]]></category>
            <category><![CDATA[data-engineering]]></category>
            <category><![CDATA[apache-iceberg]]></category>
            <dc:creator><![CDATA[Amina Okanovic]]></dc:creator>
            <pubDate>Mon, 21 Sep 2026 18:21:22 GMT</pubDate>
            <atom:updated>2026-09-21T18:21:51.250Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[PostgreSQL Will Let You Invent a Type, Right Up Until It Asks for a C Compiler]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@kabirbakovic/postgresql-will-let-you-invent-a-type-right-up-until-it-asks-for-a-c-compiler-67593aede3d6?source=rss-84c59f25e34d------2"><img src="https://cdn-images-1.medium.com/max/879/1*ju-0DYNZSRpLcX9f8tQO_Q.png" width="879"></a></p><p class="medium-feed-snippet">A partner feed lands a million order rows a day, and the order date arrives as 03/15/2021. Someone stores it in a text column because that&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@kabirbakovic/postgresql-will-let-you-invent-a-type-right-up-until-it-asks-for-a-c-compiler-67593aede3d6?source=rss-84c59f25e34d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@kabirbakovic/postgresql-will-let-you-invent-a-type-right-up-until-it-asks-for-a-c-compiler-67593aede3d6?source=rss-84c59f25e34d------2</link>
            <guid isPermaLink="false">https://medium.com/p/67593aede3d6</guid>
            <category><![CDATA[data-engineering]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[compilers]]></category>
            <category><![CDATA[database]]></category>
            <category><![CDATA[postgresql]]></category>
            <dc:creator><![CDATA[Amina Okanovic]]></dc:creator>
            <pubDate>Sun, 20 Sep 2026 20:06:21 GMT</pubDate>
            <atom:updated>2026-09-20T20:06:21.021Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[The Compact Model Was the Slow One: Storing Sparse Device Capabilities for Analytics]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@kabirbakovic/the-compact-model-was-the-slow-one-storing-sparse-device-capabilities-for-analytics-ae19b140cf30?source=rss-84c59f25e34d------2"><img src="https://cdn-images-1.medium.com/max/840/1*q3dtwOZPwrBl8mewLe3f-g.png" width="840"></a></p><p class="medium-feed-snippet">A streaming platform wants to answer one question before it ships a feature: how many devices out there can actually run it? Spatial audio&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@kabirbakovic/the-compact-model-was-the-slow-one-storing-sparse-device-capabilities-for-analytics-ae19b140cf30?source=rss-84c59f25e34d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@kabirbakovic/the-compact-model-was-the-slow-one-storing-sparse-device-capabilities-for-analytics-ae19b140cf30?source=rss-84c59f25e34d------2</link>
            <guid isPermaLink="false">https://medium.com/p/ae19b140cf30</guid>
            <category><![CDATA[analytics]]></category>
            <category><![CDATA[duckdb]]></category>
            <category><![CDATA[data-modeling]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[data-engineering]]></category>
            <dc:creator><![CDATA[Amina Okanovic]]></dc:creator>
            <pubDate>Sun, 20 Sep 2026 16:21:16 GMT</pubDate>
            <atom:updated>2026-09-20T16:21:16.009Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[A FLOAT Ledger Column Fails Two Different Ways, and Neither Looks Like a Bug]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://medium.com/@kabirbakovic/a-float-ledger-column-fails-two-different-ways-and-neither-looks-like-a-bug-381d659e3f1f?source=rss-84c59f25e34d------2"><img src="https://cdn-images-1.medium.com/max/946/1*wUZBtLh6HqtEzI0DNEnZbw.png" width="946"></a></p><p class="medium-feed-snippet">A payments team processing around 10 million transactions a day stores order_amount as DOUBLE PRECISION in Postgres, because it&apos;s the type&#x2026;</p><p class="medium-feed-link"><a href="https://medium.com/@kabirbakovic/a-float-ledger-column-fails-two-different-ways-and-neither-looks-like-a-bug-381d659e3f1f?source=rss-84c59f25e34d------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://medium.com/@kabirbakovic/a-float-ledger-column-fails-two-different-ways-and-neither-looks-like-a-bug-381d659e3f1f?source=rss-84c59f25e34d------2</link>
            <guid isPermaLink="false">https://medium.com/p/381d659e3f1f</guid>
            <category><![CDATA[data]]></category>
            <category><![CDATA[data-engineering]]></category>
            <category><![CDATA[database]]></category>
            <category><![CDATA[column]]></category>
            <category><![CDATA[database-management]]></category>
            <dc:creator><![CDATA[Amina Okanovic]]></dc:creator>
            <pubDate>Sat, 19 Sep 2026 18:20:17 GMT</pubDate>
            <atom:updated>2026-09-19T18:20:17.205Z</atom:updated>
        </item>
    </channel>
</rss>