From the team

Insights

Engineering perspective on AI, FinTech, Blockchain, and Cloud.
No fluff. No hype. Just depth.

Featured

MLOps in 2026: The Stack That Gets Models to Production Without Drama

ML models that never reach production are experiments, not products. MLOps is the engineering discipline that closes the gap between a model that works in a notebook and one that runs reliably in production for months without silent degradation. This post covers the 2026 stack we use and why each piece earns its place.

Codecanis Admin

Mar 4, 2026

9 min read Read →

All articles

Airtable

Airtable as a Backend: When It Works, When It Doesn't

Airtable feels like magic until you hit 50,000 rows or 5 requests per second. We use Airtable for plenty of production t...

Codecanis Admin

Mar 4, 2026

Consultancy

What an Architecture Audit Actually Looks Like (And When You Need One)

Companies usually ask for an architecture audit at one of three moments: before funding, before scaling, or after a near...

Codecanis Admin

Feb 26, 2026

Hiring

Hiring a Remote Project Manager: Fractional, Full-Time, or Embedded?

Engineering capacity isn't the only thing you can rent. A fractional or embedded project manager — running standups, unb...

Codecanis Admin

Feb 26, 2026

Odoo

Migrating from Odoo Community to Enterprise: What Actually Changes

We've migrated three companies from Odoo Community to Enterprise in the last 18 months. Here's what actually changes — t...

Codecanis Admin

Feb 19, 2026

Salesforce

Salesforce Apex vs HubSpot Workflows: Picking Your Custom Logic Layer

Both Salesforce and HubSpot give you a way to run custom logic, but the constraints couldn't be more different. This pos...

Codecanis Admin

Feb 12, 2026

Kubernetes

Cutting Kubernetes Costs by 60%: A Production SRE Playbook

Cloud bills are a symptom, not a root cause. After auditing Kubernetes clusters across a dozen production workloads, we'...

Codecanis Admin

Feb 11, 2026

n8n

n8n at Production Scale: Workflows That Don't Break at 3 AM

n8n is amazing for the first 100 workflows. After that, the things that worked at small scale — single-instance deployme...

Codecanis Admin

Feb 5, 2026

10m

MCP

Model Context Protocol Explained: Why MCP Is the Standard for AI Tooling

Until MCP, every AI integration was bespoke — custom adapters, custom auth, custom error handling. Model Context Protoco...

Codecanis Admin

Jan 29, 2026

Engineering digest

New articles on AI engineering, FinTech architecture, and Blockchain infrastructure — delivered when they matter, not on a schedule.

No spam. Unsubscribe any time.

Insights

MLOps in 2026: The Stack That Gets Models to Production Without Drama

All articles

Airtable as a Backend: When It Works, When It Doesn't

What an Architecture Audit Actually Looks Like (And When You Need One)

Hiring a Remote Project Manager: Fractional, Full-Time, or Embedded?

Migrating from Odoo Community to Enterprise: What Actually Changes

Salesforce Apex vs HubSpot Workflows: Picking Your Custom Logic Layer

Cutting Kubernetes Costs by 60%: A Production SRE Playbook

n8n at Production Scale: Workflows That Don't Break at 3 AM

Model Context Protocol Explained: Why MCP Is the Standard for AI Tooling

Stay sharp.