Architecture and delivery
From idea to production: requirements, APIs, data, integrations, failures, maintenance and real trade-offs.
Course author and curator
Nello Castellano
Senior Software Architect · Applied AI · Product Engineering
I have designed software for more than twenty-five years. Today I work where architecture, product and applied AI meet: systems that must do more than appear intelligent—they must work, explain themselves and withstand verification.
- Based in
- Singapore
- Experience
- 25+ years
- Focus
- Applied AI and product engineering
The model matters, but it is not enough on its own. Value comes from what we build around it: context, tools, boundaries, controls and accountability.
That idea guides this course. It is not a collection of prompts to copy, but a way of working that makes every step observable and every conclusion supported by evidence.
The path
From software that must work to AI that must prove
I have worked in software since 2001 and am now based in Singapore. I build enterprise platforms, SaaS applications, cloud infrastructure, databases and APIs. The serious work remains the same: understand the problem, design a sustainable solution, find where it breaks and take it to production.
In recent years I have brought that discipline into applied AI. I use Large Language Models, Claude Code and OpenAI Codex for implementation, refactoring, documentation and testing. Architecture, integration, evaluation and final acceptance remain human responsibilities.
Agentic Workshop grew from that experience: explain concepts from first principles, then make them real through assignments, simulations and controls. If an idea cannot survive an example, a realistic failure and a verification step, it is not yet a lesson.
What I bring to the course
Three disciplines working together
Applied AI
LLMs, structured outputs, evaluation, automation and human-in-the-loop controls built into a system rather than left in a demo.
Operational explanation
Minimal jargon, distinct examples, visible consequences and evidence that shows where theory ends.
Field experience
Projects that shaped the method
These are not fictional case studies. They are products and systems I have worked on, with real architecture, user-experience, verification and release constraints.
XTREAM IPTV PLAYER PRO+
A native tvOS player for user-authorised television services, with separate profiles, very large catalogues, local search, programme guide and catalogue-aware recommendations.
Open in the App Store ↗PDFMicroSuite
A native Mac app to merge, split, compress, annotate and protect PDFs. Document work stays local, while AI features let users choose a provider.
Visit the official site ↗API Harness
An API-first system for LLM-assisted psychometric scoring: versioned evidence-based assessments, controlled scoring and human review before sensitive decisions.
Read the project context ↗OtterForm
A platform for forms and conversational experiences with AI-assisted generation, conditional logic, analytics, video and integrations.
Visit the product ↗QR-Code Scanner and Generator
A fast app for scanning QR codes and barcodes from the camera or images and generating more than fifteen code types, with no ads or account.
Open in the App Store ↗Pizza Calculator Pro
A dough calculator combining style, hydration, temperature, proofing time, dough-ball weight and preferments; it saves recipes and works offline.
Open in the App Store ↗Experience
A long technical foundation, not an overnight credential
- Since 2001
- Professional software experienceToday: Senior Software Engineer / Software Architect, based in Singapore.
- 2019–2020
- Specialist Diploma in Full Stack Web DevelopmentSingapore Polytechnic.
- Continuous learning
- Applied and Agentic AILLM integration, MCP, RAG, structured output, evaluation, workflow automation and human-in-the-loop systems.
Want to see this method at work?
Start with the foundations or open exactly the module you need. The path is free and never locks you into a mandatory sequence.
Start the workshop