<?xml version="1.0" encoding="UTF-8" ?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Navik Engineering Blog</title>
    <link>https://navik.dev/blog/</link>
    <description>Field notes from building AI agents, data pipelines, quant systems and MVPs - what broke, why, and how we fixed it.</description>
    <language>en-us</language>
    <lastBuildDate>Tue, 17 Nov 2026 07:22:38 GMT</lastBuildDate>
    <atom:link href="https://navik.dev/feed.xml" rel="self" type="application/rss+xml" />
    <item>
      <title><![CDATA[Choosing a database in 2026: a consultant's cheat sheet]]></title>
      <link>https://navik.dev/blog/choosing-a-database-2026/</link>
      <guid isPermaLink="true">https://navik.dev/blog/choosing-a-database-2026/</guid>
      <pubDate>Thu, 22 Jan 2026 00:00:00 GMT</pubDate>
      <category><![CDATA[Data Engineering]]></category>
      <description><![CDATA[Supabase, Postgres, ClickHouse - we get asked which one constantly. The honest answer is that they mostly are not competitors, and the decision is easier than it looks once you ask the right question first.]]></description>
    </item>
    <item>
      <title><![CDATA[Putting Claude into production without it embarrassing you]]></title>
      <link>https://navik.dev/blog/claude-in-production/</link>
      <guid isPermaLink="true">https://navik.dev/blog/claude-in-production/</guid>
      <pubDate>Thu, 20 Nov 2025 00:00:00 GMT</pubDate>
      <category><![CDATA[AI & Agents]]></category>
      <description><![CDATA[An LLM in a demo and an LLM in a feature real users depend on are different engineering problems. Here is how we put Claude into production systems so that it is dependable, not just impressive.]]></description>
    </item>
    <item>
      <title><![CDATA[What 'minimum' actually means in MVP]]></title>
      <link>https://navik.dev/blog/mvp-scope-what-minimum-means/</link>
      <guid isPermaLink="true">https://navik.dev/blog/mvp-scope-what-minimum-means/</guid>
      <pubDate>Thu, 18 Sep 2025 00:00:00 GMT</pubDate>
      <category><![CDATA[MVP & Product]]></category>
      <description><![CDATA[Most MVPs are not minimum and not viable. After enough of them, we have learned that scoping an MVP is the hard part - and that almost every instinct a founder has about it points the wrong way.]]></description>
    </item>
    <item>
      <title><![CDATA[Idempotency, retries and backfills: pipelines you can trust]]></title>
      <link>https://navik.dev/blog/idempotent-data-pipelines/</link>
      <guid isPermaLink="true">https://navik.dev/blog/idempotent-data-pipelines/</guid>
      <pubDate>Thu, 14 Aug 2025 00:00:00 GMT</pubDate>
      <category><![CDATA[Data Engineering]]></category>
      <description><![CDATA[Every data pipeline we have built eventually taught us the same three lessons. They are not advanced. They are just the difference between a pipeline you can rerun without fear and one you cannot.]]></description>
    </item>
    <item>
      <title><![CDATA[The two stacks we actually use]]></title>
      <link>https://navik.dev/blog/two-stacks-mvp-vs-enterprise/</link>
      <guid isPermaLink="true">https://navik.dev/blog/two-stacks-mvp-vs-enterprise/</guid>
      <pubDate>Tue, 08 Jul 2025 00:00:00 GMT</pubDate>
      <category><![CDATA[Cloud & Architecture]]></category>
      <description><![CDATA[After three years and somewhere north of thirty projects, we have stopped picking a stack from scratch each time. We have two, and most of the work is being honest about which one a project actually needs.]]></description>
    </item>
    <item>
      <title><![CDATA[The Postgres patterns we keep coming back to]]></title>
      <link>https://navik.dev/blog/postgres-patterns-long-term-projects/</link>
      <guid isPermaLink="true">https://navik.dev/blog/postgres-patterns-long-term-projects/</guid>
      <pubDate>Thu, 12 Jun 2025 00:00:00 GMT</pubDate>
      <category><![CDATA[Data Engineering]]></category>
      <description><![CDATA[Across years of long-term client projects, the same handful of Postgres habits keep proving their worth. None of them are clever. All of them save you from a bad afternoon later.]]></description>
    </item>
    <item>
      <title><![CDATA[When we reach for ClickHouse instead of Postgres]]></title>
      <link>https://navik.dev/blog/clickhouse-vs-postgres-analytics/</link>
      <guid isPermaLink="true">https://navik.dev/blog/clickhouse-vs-postgres-analytics/</guid>
      <pubDate>Thu, 15 May 2025 00:00:00 GMT</pubDate>
      <category><![CDATA[Data Engineering]]></category>
      <description><![CDATA[Postgres is our default for almost everything. ClickHouse is not a replacement for it - it is the answer to a specific question, and knowing which question matters more than knowing either database.]]></description>
    </item>
    <item>
      <title><![CDATA[A full AWS and Postgres build, end to end]]></title>
      <link>https://navik.dev/blog/supermatches-aws-postgres-build/</link>
      <guid isPermaLink="true">https://navik.dev/blog/supermatches-aws-postgres-build/</guid>
      <pubDate>Thu, 10 Apr 2025 00:00:00 GMT</pubDate>
      <category><![CDATA[Cloud & Architecture]]></category>
      <description><![CDATA[Supermatches.in was not a one-week MVP. It was a full build on AWS with Postgres at the core - the other end of the spectrum from managed-everything, and a good lesson in when that trade is worth making.]]></description>
    </item>
    <item>
      <title><![CDATA[Shipping a real MVP in a week with Supabase and Vercel]]></title>
      <link>https://navik.dev/blog/supabase-vercel-one-week-mvp/</link>
      <guid isPermaLink="true">https://navik.dev/blog/supabase-vercel-one-week-mvp/</guid>
      <pubDate>Thu, 13 Mar 2025 00:00:00 GMT</pubDate>
      <category><![CDATA[MVP & Product]]></category>
      <description><![CDATA[Hangout.in, by UnboxingBLR, went from idea to live product in about a week. That is only possible because of what we deliberately did not build - and Supabase plus Vercel let us not build a lot.]]></description>
    </item>
    <item>
      <title><![CDATA[When one backtest's output is another's input]]></title>
      <link>https://navik.dev/blog/chaining-dependent-strategies/</link>
      <guid isPermaLink="true">https://navik.dev/blog/chaining-dependent-strategies/</guid>
      <pubDate>Thu, 20 Feb 2025 00:00:00 GMT</pubDate>
      <category><![CDATA[Quant & Fintech]]></category>
      <description><![CDATA[Single strategies are tidy. The moment one strategy consumes another's output, you have a pipeline - and on Weatherstone, managing that web of dependencies became the difference between trustworthy results and a tangle.]]></description>
    </item>
    <item>
      <title><![CDATA[Historical data gets revised. Your backtest doesn't know that.]]></title>
      <link>https://navik.dev/blog/historical-data-revisions/</link>
      <guid isPermaLink="true">https://navik.dev/blog/historical-data-revisions/</guid>
      <pubDate>Thu, 16 Jan 2025 00:00:00 GMT</pubDate>
      <category><![CDATA[Quant & Fintech]]></category>
      <description><![CDATA[You backtest a strategy, get a result, and move on. Months later you pull the same history and the numbers have changed - because history changed. On Weatherstone, designing for a past that is not fixed was its own problem.]]></description>
    </item>
    <item>
      <title><![CDATA[Backtesting on free market data: garbage in, confident garbage out]]></title>
      <link>https://navik.dev/blog/backtesting-free-market-data/</link>
      <guid isPermaLink="true">https://navik.dev/blog/backtesting-free-market-data/</guid>
      <pubDate>Thu, 14 Nov 2024 00:00:00 GMT</pubDate>
      <category><![CDATA[Quant & Fintech]]></category>
      <description><![CDATA[For Weatherstone's quant work, the strategies were only ever as trustworthy as the data underneath them. Free market data is full of quiet problems, and a backtest will happily turn those problems into a number that looks like a result.]]></description>
    </item>
    <item>
      <title><![CDATA[Your sample data is lying to you]]></title>
      <link>https://navik.dev/blog/sample-data-lies/</link>
      <guid isPermaLink="true">https://navik.dev/blog/sample-data-lies/</guid>
      <pubDate>Thu, 10 Oct 2024 00:00:00 GMT</pubDate>
      <category><![CDATA[Data Engineering]]></category>
      <description><![CDATA[On a terabyte-scale migration for Candu, the sample data looked clean and the full dataset did not. The bug you find in production is the cheap one - the data-quality bug you find three hours into a run is the expensive one.]]></description>
    </item>
    <item>
      <title><![CDATA[When your database bill grows faster than your data]]></title>
      <link>https://navik.dev/blog/database-storage-costs/</link>
      <guid isPermaLink="true">https://navik.dev/blog/database-storage-costs/</guid>
      <pubDate>Thu, 19 Sep 2024 00:00:00 GMT</pubDate>
      <category><![CDATA[Data Engineering]]></category>
      <description><![CDATA[An IoT platform generates data forever and deletes none of it. On Sustainergy we watched storage costs climb until we had to get serious about what data is worth keeping hot, and what is not.]]></description>
    </item>
    <item>
      <title><![CDATA[Thousands of calculations a second, and what happens when one fails]]></title>
      <link>https://navik.dev/blog/dependent-trigger-calculations/</link>
      <guid isPermaLink="true">https://navik.dev/blog/dependent-trigger-calculations/</guid>
      <pubDate>Thu, 15 Aug 2024 00:00:00 GMT</pubDate>
      <category><![CDATA[Data Engineering]]></category>
      <description><![CDATA[Sustainergy turns raw sensor data into insights through a web of calculations that depend on each other. Designing for the failure of any single one - without taking down the rest - was the real work.]]></description>
    </item>
    <item>
      <title><![CDATA[Scraping the internet for a content engine, without it quietly falling over]]></title>
      <link>https://navik.dev/blog/pockla-scraping-pipeline/</link>
      <guid isPermaLink="true">https://navik.dev/blog/pockla-scraping-pipeline/</guid>
      <pubDate>Thu, 25 Jul 2024 00:00:00 GMT</pubDate>
      <category><![CDATA[Data Engineering]]></category>
      <description><![CDATA[Behind Pockla's AI is a scraping pipeline that has to keep a content pool full from sources that change, rate-limit and break constantly. Celery, RabbitMQ, and a long list of lessons about backpressure.]]></description>
    </item>
    <item>
      <title><![CDATA[Building a content pool you can actually generate from]]></title>
      <link>https://navik.dev/blog/pockla-content-pool-rag/</link>
      <guid isPermaLink="true">https://navik.dev/blog/pockla-content-pool-rag/</guid>
      <pubDate>Thu, 13 Jun 2024 00:00:00 GMT</pubDate>
      <category><![CDATA[AI & Agents]]></category>
      <description><![CDATA[Pockla's content engine runs on a pool of scraped, processed content. Keeping that pool fresh, relevant and small enough to query fast turned into the real engineering problem behind the AI.]]></description>
    </item>
    <item>
      <title><![CDATA[Teaching an agent what a brand actually sounds like]]></title>
      <link>https://navik.dev/blog/pockla-brand-profiling-agent/</link>
      <guid isPermaLink="true">https://navik.dev/blog/pockla-brand-profiling-agent/</guid>
      <pubDate>Thu, 09 May 2024 00:00:00 GMT</pubDate>
      <category><![CDATA[AI & Agents]]></category>
      <description><![CDATA[A brand profile is easy to describe and hard to build. For Pockla we needed an agent that could read everything a brand had ever published and come back with something a content engine could use. Here is what that took.]]></description>
    </item>
    <item>
      <title><![CDATA[Trending is easy. On-brand and usable is the hard part.]]></title>
      <link>https://navik.dev/blog/pockla-trending-content-relevance/</link>
      <guid isPermaLink="true">https://navik.dev/blog/pockla-trending-content-relevance/</guid>
      <pubDate>Thu, 18 Apr 2024 00:00:00 GMT</pubDate>
      <category><![CDATA[AI & Agents]]></category>
      <description><![CDATA[For Pockla we had to surface trending moments for brands. The catch: most trends are either irrelevant to the brand or impossible to make content from. Here is how we taught the system to tell the difference.]]></description>
    </item>
    <item>
      <title><![CDATA[Ingesting thousands of IoT sensors without losing data]]></title>
      <link>https://navik.dev/blog/iot-thousands-of-sensors-ingestion/</link>
      <guid isPermaLink="true">https://navik.dev/blog/iot-thousands-of-sensors-ingestion/</guid>
      <pubDate>Thu, 07 Mar 2024 00:00:00 GMT</pubDate>
      <category><![CDATA[Data Engineering]]></category>
      <description><![CDATA[Sustainergy runs on data from thousands of energy sensors reporting continuously. The hard part was never the volume - it was the gaps, the duplicates, and the sensors that lie. Lessons from the ingestion layer.]]></description>
    </item>
    <item>
      <title><![CDATA[Terabytes on AWS Glue: every failed run costs you twice]]></title>
      <link>https://navik.dev/blog/glue-tbs-of-data-pitfalls/</link>
      <guid isPermaLink="true">https://navik.dev/blog/glue-tbs-of-data-pitfalls/</guid>
      <pubDate>Thu, 22 Feb 2024 00:00:00 GMT</pubDate>
      <category><![CDATA[Data Engineering]]></category>
      <description><![CDATA[Migrating multiple terabytes through AWS Glue sounds like a configuration task. It is not. When a run takes hours and a failure near the end means starting over, the economics change how you have to build.]]></description>
    </item>
    <item>
      <title><![CDATA[What building custom AI agents actually looks like]]></title>
      <link>https://navik.dev/blog/custom-ai-agents/</link>
      <guid isPermaLink="true">https://navik.dev/blog/custom-ai-agents/</guid>
      <pubDate>Wed, 10 Jan 2024 00:00:00 GMT</pubDate>
      <category><![CDATA[AI & Agents]]></category>
      <description><![CDATA[Most writing about AI agents is about the demo. This is about the other 90% - the boundaries, the validation, the decomposition and the unglamorous engineering that decides whether an agent survives contact with production.]]></description>
    </item>
    <item>
      <title><![CDATA[Why early-stage startups bring in a fractional CTO]]></title>
      <link>https://navik.dev/blog/fractional-cto-benefits/</link>
      <guid isPermaLink="true">https://navik.dev/blog/fractional-cto-benefits/</guid>
      <pubDate>Wed, 20 Dec 2023 00:00:00 GMT</pubDate>
      <category><![CDATA[Leadership]]></category>
      <description><![CDATA[A fractional CTO is not a cheaper full-time CTO. It is a different thing entirely, and it is the right thing for a specific, common situation that a lot of early founders find themselves in.]]></description>
    </item>
  </channel>
</rss>
