Blog
Notes from the build
Practical engineering writing — the things we learn shipping Databricks platforms, Retool tools and Django products for clients.
Domo: what it actually does, and where it earns its keep
Domo bundles data connectors, low-code ETL and dashboards into one cloud platform. Here's what that buys you in practice, and where it runs out of road.
Read articleShipping a data app on the lakehouse: Databricks Apps in practice
Databricks Apps hosts interactive tools inside the workspace, with Unity Catalog access and auth built in. Where that's a genuine win, and where Retool still wins.
Read articleWhat the Databricks REST API can and can't automate in a large migration
Bulk-exporting jobs, clusters and permissions via API is the easy 80%. The hard 20% of a large Databricks migration is judgment the API cannot make for you.
Read articleDatabricks SQL vs a classic BI warehouse: what actually changes for your analysts
DBSQL looks like just another warehouse endpoint from a BI tool. For analysts day to day, several real things change — and a few important ones don't.
Read articleMosaic AI for a first GenAI feature: what has to be in place before model serving
Standing up a Mosaic AI serving endpoint takes an afternoon. Answer quality depends on the retrieval, evaluation and governance work that has to happen first.
Read articleServerless Databricks: when it's cheaper and when it isn't
Serverless removes cluster ops and idle spend, but it carries a per-DBU premium over classic compute. Here's the actual usage pattern that decides which one wins.
Read articleUnity Catalog and open formats: Delta, Iceberg and UniForm without picking a side
You don't have to choose between Delta and Iceberg. UniForm and Unity Catalog's Iceberg REST endpoint let you write once and serve both — here's how, and where the seams still show.
Read articleSnowflake to Databricks: what the migration actually costs you
A phased playbook for moving off Snowflake — where the effort really goes, why SQL dialect debt dominates the timeline, and how to model breakeven honestly.
Read articleAgentic AI and big data: what your platform needs before agents are useful
Agents that act on enterprise data need more than a good model — governed tables, narrow tools, real permissions and a cost ceiling. What has to be true first.
Read articleHiring Databricks-Certified Data Engineers: What the Certification Tells You
A Databricks certification proves platform fluency, not architecture judgment. What it actually verifies, and how to interview past the badge.
Read articleDatabricks cost audit: where clusters quietly burn money
The usual waste patterns behind a growing Databricks bill, what a real cost audit checks, and realistic savings ranges once you fix them.
Read articleDatabricks Genie: what natural-language BI actually gets right
A level-headed look at Databricks Genie: where it removes analyst workload, what setup it needs, and where it still falls short.
Read articleAutomating Databricks with the REST API: A Field Guide
What's actually worth automating on Databricks, when to use the REST API vs Terraform, and how to avoid the auth, rate-limit and drift mistakes teams make first.
Read articleDatabricks Workflows vs Airflow vs Dagster: Picking an Orchestrator
Databricks Workflows, Airflow, and Dagster solve orchestration differently. A practical decision framework based on your stack, not a vendor pitch.
Read articleDelta Live Tables in practice: what it saves you, where it gets in the way
An honest look at Delta Live Tables versus hand-rolled Spark pipelines: what DLT genuinely saves, where it costs you control, and when to skip it.
Read articleHive Metastore vs Unity Catalog: what actually changes
A concrete comparison of Hive metastore and Unity Catalog on Databricks: permissions, lineage, cross-workspace scope, and what breaks operationally.
Read articleProductionizing ML on Databricks with MLflow
MLflow solves experiment tracking and model versioning. It does not solve drift, retraining, or feature consistency. Here's the realistic path from notebook to production.
Read articleUnity Catalog governance: a practical playbook
You have Unity Catalog. Now what? Naming, GRANT design, row filters and column masks, lineage for audits, tagging, and the failure modes to avoid.
Read articleWhy hire a Databricks Consulting Partner instead of building in-house
Zephico is now a Databricks Consulting Partner. Here's the honest tradeoff between building Databricks capability in-house and bringing in a partner.
Read articleMigrating to Unity Catalog: what it actually takes
Unity Catalog is no longer optional on Databricks. Here's what the migration from Hive metastore really involves, where it hurts, and how long it takes.
Read articleContract engineer vs. full-time hire: the real cost math
Salary is less than half the story. A line-by-line look at what a senior engineer actually costs to recruit, employ and — sometimes — let go.
Read articleHow much does custom software cost in 2026? An honest breakdown
Agencies hate publishing prices. Here are the real cost drivers, realistic ranges, and the questions that keep a project on the right side of them.
Read articleDatabricks or Snowflake? How mid-size companies should choose
The platforms have converged on paper. The right choice still comes down to your workloads, your team, and how each platform bills you for growth.
Read articleLLM features: the distance between the demo and production
The demo takes a week and impresses everyone. What separates it from a feature you can ship — retrieval, evals, guardrails and cost — is the actual project.
Read articleOutsourcing development under GDPR: what to demand from your vendor
A buyer's checklist for hiring an offshore development partner without creating a compliance problem — DPAs, transfers, access control and the questions that expose weak answers.
Read articleSaleor vs Shopify Plus: when owning your stack starts paying off
Shopify wins the first years for almost everyone. The interesting question is what happens after — and when open-source headless commerce flips the math.
Read articleWhat an internal tool on Retool actually costs
License, build, and the maintenance nobody budgets for — the full cost picture of a Retool project, compared honestly against the spreadsheet it replaces.
Read articleWhere AWS savings actually hide: notes from cost audits
Most cloud bills carry 20–40% of pure waste, and it clusters in the same places. A field guide, ordered by effort-to-savings ratio.
Read articleMedallion architecture on Databricks: what actually matters
Bronze, silver, gold — everyone draws the same diagram. Here's what we've learned actually matters when you build a medallion lakehouse for production.
Read articleWhen Retool is the right answer (and when it isn't)
We build internal tools with Retool for a living — which also means we know exactly where its limits are. A practical decision guide.
Read articleHow to make contract engineers actually work: lessons from the other side
We place engineers into North American and European teams for a living. Here's what separates the engagements that compound from the ones that stall.
Read articleNeed engineers next week, not next quarter?
Tell us what you are building and the skills you are missing. We will come back within one business day with profiles and a plan.