Search
Explore digital transformation resources.
Uncover insights, best practises and case studies.
Search
Explore digital transformation resources.
Uncover insights, best practises and case studies.
Tech talk
Oct 14 (Wed) | 19:00 EEST time | Nortal Bulgaria office
AI coding agents are becoming part of everyday software development. Using them well means understanding how they work, knowing where they help, questioning their output when needed and staying in control.
In this tech talk, Ivan Mihov, senior software engineer in the News UK Tech dedicated team at Nortal Bulgaria, will share his approach to working with coding agents, focusing on finding the right balance between development speed and engineering judgment. He will look beyond the hype around vibe coding and explore a more structured way of using AI in software development.
Ivan will introduce Claude Code, one of the most discussed AI coding tools today, and explain how AI agents work as software systems. Attendees will also see an example of a small project built using a spec-driven workflow where the engineer remains responsible for guiding and validating every step.
Validation is where engineering judgment matters most. During the presentation, Ivan will show why human review remains essential, both when using AI coding systems and in broader AI applications. He'll offer a practical engineering perspective on how to use AI tools with a clear sense of ownership over the result.
Join us for an in-depth discussion on building software with AI coding agents, what works today, where the limits are, and how to use these tools with confidence while staying responsible for the outcome.
18:30 – 19:00 Doors open
19:00 – 20:30 Presentation
20:30 – 21:00 Q&A and discussion
Ivan will open the talk by drawing a distinction he sees as fundamental: vibe coding, where users lean entirely on the agent's suggestions, versus agentic engineering, where a developer directs it with technical judgment and checks its work at every step. The gap between the two decides whether a coding agent speeds up good engineering, or just speeds up bad code reaching production.
Many developers use coding agents daily without stopping to consider how they actually work: as software, with their own logic and dependencies. You'll get an engineer's breakdown of what's actually happening and how that changes the way you prompt.
You'll watch Ivan build a small project live with Claude Code, currently one of the most widely used tools for this kind of work. He'll walk through his own spec-driven process, settling the specification first, so every change the agent makes can be checked against it, rather than reviewed after the fact.
Validation comes down to human-in-the-loop, a concept familiar from many AI systems beyond coding. Ivan will connect that to a wider pattern across those systems, a link he's drawn from his own work rather than something he's read, and explain why skipping that check tends to cost more than it saves.
If you're a software engineer looking to grow your skills, this talk gives you a practical, disciplined way to work with coding agents.
Skepticism is fair. This talk makes the case for a more deliberate way of using coding agents, not just the hype.
Tech leads carry the call on how a team adopts coding agents. This talk gives you a concrete, accountable example to work from.
If you work close to engineering teams and need to understand how coding agents affect delivery and quality, this talk offers a clear perspective on what responsible adoption looks like.
Ivan Mihov is a senior software engineer at Nortal with over a decade of experience building enterprise platforms for trading, e-commerce and private cloud automations. In his current role, Ivan works on the customer and subscription platforms for the digital edition of the British newspaper The Sun.
He is a highly versatile developer whose expertise spans Java/SpringBoot and Node.js, with additional hands-on experience in .NET/C#, Python and beyond.
As an early pioneer in AI-assisted development, Ivan has integrated AI into his engineering workflow since the release of ChatGPT in late 2022. He has extensive, hands-on experience with the entire ecosystem of major AI coding tools. He has used tools like Claude Code, GitHub Copilot, Codex and Cursor to tackle complex tasks across various technology stacks.
A few articles from the speaker on topics beyond this talk.
Seats are limited.
Register now to join us at the Nortal Bulgaria office.