Deep Engineering #23: Frameworks as Flow Control with Ivan Padabed and Roman Voronin
What leaders should ask before picking an ADF—and how IBM’s 2025–26 trends light the path forward
Build your next app on HubSpot with the flexibility of an all-new Developer Platform
The HubSpot Developer Platform gives you the tools to build, extend, and scale with confidence. Create AI-ready apps, integrations, and workflows faster with a unified platform designed to grow alongside your business.
✍️From the editor’s desk,
IBM Think’s trend brief published at the start of this month puts application development at the center of digital transformation in 2025–26: AI/ML is being woven through the lifecycle; low/no-code is accelerating toward a projected USD 264B by 2032; cloud-native DevOps is shifting to DevSecOps by default; and IoT and PWAs are reshaping delivery and UX across sectors.
In this context, how should engineering leaders choose and operationalize an Application Development Framework (ADF) to increase team throughput without fragmenting architecture?
To find the answer in this issue, we’ve collaborating with the Ivan Padabed and Roman Voronin, authors of Building an Application Development Framework (Packt, 2025). Padabed is co-founder and CEO at System5Dev, a startup focused on applying AI to systems architecture; Cloud Platform Architect at Intapp (a B2B SaaS platform); an Expert at Primary Venture Mastermind Network: a 25+-year systems-architecture leader, educator, and community organizer. He’s also an active researcher in systems architecture and AI. Voronin is the co-founder and CTO of System5Dev, where he shapes AI-ready system designs, and a Senior MLOps Engineer at Intapp. With more than twenty years in technology, he champions AWS-driven DevOps, shares insights as a long-standing AWS Community Builder, and authors practical articles on cloud architecture.
Highlights from today’s issue:
Feature Article: Frameworks as Flow Control with Ivan Padabed and Roman Voronin
The complete “Chapter 1: Introduction to Application Development Frameworks” from the book, Building an Application Development Framework (Packt, 2025), by Ivan Padabed and Roman Voronin
5 application development trends to watch in 2025 and 2026 by Mesh Flinders | IBM Think
Let’s get started!
Sponsored:
Ship Faster with DevStats: Spot bottlenecks early, keep engineers in flow, and deliver up to 30% faster.
Frameworks as Flow Control with Ivan Padabed and Roman Voronin
In Building an Application Development Framework, Ivan Padabed and Roman Voronin define an ADF as:
“a collection of pre-written code and tools that provide a structured approach to building applications,”
distinguishing it from libraries and SDKs by its architectural enforcement and inversion of control.
Unlike libraries, which you call, frameworks call you. This inversion of control is what allows ADFs to impose a consistent structure and flow—ensuring teams follow standardized development practices.
The Productivity Pattern: Less Repetition, More Signal
Padabed and Voronin note that ADFs encode reusable architectural blueprints—from MVC to event-driven and middleware pipelines—so teams don’t waste energy rewriting infrastructure. This thesis is echoed in industry data. According to the 2025 Stack Overflow Developer Survey, React (used by 44.7% of respondents) and Node.js (48.7%) dominate because they abstract away boilerplate and enforce repeatable project structures across teams.
These frameworks don’t just save time—they reduce variability. In distributed systems or polyglot teams, structure becomes an asset. For example, Django enforces MVT, Spring Boot applies layered architecture, and FastAPI adopts declarative schema-first design with OpenAPI by default. These design choices encode architectural conventions directly into tooling.
ADF as a Subsystem of SDLC
Padabed and Voronin frame the ADF not as a standalone artifact, but as a subsystem of the broader software development lifecycle (SDLC). This is key. Frameworks operate most effectively when they’re tightly integrated into iterative delivery models: testing, linting, CI/CD, and observability hooks are all structured around framework conventions. For example, Next.js’s file-system routing pairs naturally with Vercel’s deployment pipeline, and Angular’s CLI scaffolds modules aligned with end-to-end testing via Protractor or Cypress.
This integration is what makes ADFs strategic. They not only reduce developer cognitive load but improve team-level throughput. When every project in an org shares the same conventions—controller locations, dependency injection syntax, build pipelines—handoffs get faster, onboarding is easier, and architectural drift is minimized.
Trends: The Framework Extends
In past few months we have seen three key trends emerge in how frameworks are extending beyond their traditional domains:
FastAPI Gains Ground: The 2025 Stack Overflow survey shows FastAPI gaining +5 percentage points in web framework usage—the largest jump across any backend tech. Its combination of speed, type hints, and auto-generated docs has made it the preferred choice for Python API services
AI Frameworks Enter the ADF Space: Frameworks like LangChain (33% adoption among AI developers) and Ollama (51%) are increasingly used to build agentic systems. These aren’t just wrappers—they impose lifecycle patterns (e.g. chains, memory, retrieval-augmented generation) that formalize how AI apps interact with models and data.
Low-Code Frameworks Mature: The low-code space has moved well beyond internal tooling. Microsoft’s Power Platform and Mendix are now being embedded in enterprise SDLCs as sanctioned ADFs for line-of-business automation. According to IBM, the global market for low-code development is projected to reach $264 billion by 2032, with frameworks driving the shift from form builders to orchestrated, policy-aware app stacks.
Choosing an ADF Is Choosing a Direction
Engineers often focus on syntax, speed, or community when selecting a framework. However, for leaders, the more critical question is: what kind of system behavior does this framework enforce?
Frameworks impose boundaries. They encode opinions. They make certain mistakes harder and certain workflows inevitable. This is why Elixir’s Phoenix framework—used by a small fraction of developers—was rated the “most admired” framework of 2025 (79% satisfaction), despite its niche market.
The following is a checklist your team can use as a starting point:
ADF selection & rollout checklist:
Architecture & control flow fit: Does the framework’s execution model (routing, DI, background jobs, testing) match how you build and ship today (or want to)? e.g., Next.js documents zero-config, first-class deployment on Vercel that aligns routing/build with CI/CD out of the box.
CI/CD & operations alignment: Prefer frameworks with opinionated tooling (CLIs, schematics, deploy adapters) that reduce bespoke pipeline glue; Angular’s CLI scaffolding and Cypress schematic are examples that shorten e2e setup and keep tests on the happy path.
Security posture & patch cadence: Check LTS lines, security advisories, and ease of minor/major upgrades—teams that stay current realize measurable perf/cost wins (see this Netflix example).
Onboarding & consistency: Favor frameworks that enforce conventions (project layout, testing patterns) to speed new-hire time-to-first-PR and reduce variance across services; measure with DORA lead time/change failure rate.
Ecosystem & runway: Confirm a healthy plugin/library ecosystem and provider backing (docs, adapters, migration guides) so you’re not boxed in when requirements evolve.
Business alignment to trends: Map choices to near-term forces—AI/ML in the lifecycle, rising low/no-code, cloud-native DevSecOps, IoT/PWAs—as flagged by IBM Think’s 2025–26 outlook.
If you are ready to dive deeper into ADFs, we’ve included the complete Chapter 1: Introduction to Application Development Framework from Padabed and Voronin’s book, Building an Application Development Framework. The chapter defines an Application Development Framework (ADF) as a convention-driven, inversion-of-control “skeleton” for building software; distinguishes it from APIs, libraries, SDKs, platforms, and DSLs; situates it historically and across major categories (web, enterprise, low-code, AI/ML, etc.); and positions the ADF as a subsystem of the SDLC—valued for standardizing architecture and flow to reduce cognitive load, improve maintainability, and boost team throughput.
🧠Expert Insight
🛠️Tool of the Week
Hono (v4.10.x) — a tiny, blister-fast, open-source web framework that runs the same app across Workers/edge (Cloudflare, Fastly), Node.js, Deno, Bun, and Vercel with first-class TypeScript.
Highlights:
Proven at the edge & in prod: Official Cloudflare docs recommend Hono for Workers/Pages, and Cloudflare’s own blog lists companies (e.g., Unkey, Nodecraft) using it in production.
Modern, portable stack: One codebase deploys to multiple runtimes with “batteries-included” middleware and a clean TS API; released under MIT.
📎Tech Briefs
5 application development trends to watch in 2025 and 2026 by Mesh Flinders | IBM Think: Says 2025–26 app development will be defined by AI/ML integration, the rise of low/no-code, cloud-native DevOps with DevSecOps, expanding IoT/edge and PWAs—improving UX, automation, security, and sector outcomes—with 2026 bringing more mobile-first strategies, deeper automation, gen-AI/IoT leadership, and teams oriented around DevSecOps, all driven by business needs rather than tech novelty.
PyTorch team unveils framework for programming clusters: On October 22, 2025, Meta’s PyTorch team announced Monarch, an experimental actor-based distributed programming framework—with a Python front end and Rust back end—that integrates with PyTorch to let developers program entire GPU clusters “like one machine,” handling sharding, vectorization, and GPU-to-GPU transfers while currently warning of bugs and evolving APIs.
LLM-first Web Framework - Minko Gechev | Google: Outlines an “LLM-first” approach to web frameworks and tooling, sharing experiments with a tiny framework (Revolt) and sketches a vision of agentic, browser-native workflows using MCP tools and runtime introspection to automate routine tasks.
Next.js 16 (beta) announced: Turbopack becomes the default bundler (promising major dev-loop speedups), with React Compiler support, smarter routing/prefetching, and new caching APIs.
ASP.NET Core (Kestrel) critical vuln patched: Microsoft addressed CVE-2025-55315 (CVSS 9.9) in Patch Tuesday; coverage notes it as the framework’s “highest-ever” severity to date and urges prompt updates.
That’s all for today. Thank you for reading this issue of Deep Engineering. We’re just getting started, and your feedback will help shape what comes next. Do take a moment to fill out this short survey we run monthly—as a thank-you, we’ll add one Packt credit to your account, redeemable for any book of your choice.
We’ll be back next week with more expert-led content.
Stay awesome,
Divya Anne Selvaraj
Editor-in-Chief, Deep Engineering
If your company is interested in reaching an audience of developers, software engineers, and tech decision makers, you may want to advertise with us.
 |
|
 | A guest post by
|