Tech talks from Korea's community

Finding a bug is one thing. Turning it into a reliable exploit is another, and LLM automation is changing both. Join djnn for a look at exploit engineering: what it takes to go from vulnerability research to a working exploit, and the ever-evolving challenges that come with it in the age of LLM automation.

Everyone is talking about AI, but few understand how it works. How transformers work, why hallucinations and shaky benchmarks aren't going away, how developers can build senior-level skills without letting AI erode them, and what AI is doing to society.

With AI you can build an app in an afternoon, including the wrong one. How careful planning, small apps, and a few agents that each own one job turned a rebuilt kids' education app into a $22,000 exit.

AI can make an ad in seconds, but most AI-made ads flop. An AI-assisted system for going from idea to creative: market research, personas and angles, and checking alignment from ad to paywall, with human judgment deciding what ships.

AI agents that move money face risks at every step, from forged instructions to front-running. A framework of gates, from runtime checks and signing policies to on-chain guards and builder-level protection, that keeps financial agents safe from intent to settlement.

Why China's AI ecosystem is worth a closer look for founders, through three pillars: low-cost open models, a deep pool of AI talent, and strong policy support. Three founders in residence at The Arch then share what they're building in Shanghai.

Adding a purchase button is easy; the hard part is everything after payment. How a billing platform helps subscription apps manage revenue, test paywalls, handle billing edge cases, and keep access consistent across platforms.

Why every LLM, from ChatGPT to Claude, inherits the non-determinism and hallucinations of the Transformer architecture, and how developers can adapt by accelerating their path to senior and building fundamentals that coding agents amplify instead of replace.

Why AI agents that work perfectly on your laptop fall apart in production, and how checkpointing agent state to a database keeps a killed process from burning the same tokens twice.

A look at Qwen3.8 Max as a drop in for the coding execution phase: benchmark scores against frontier models, OpenAI compatible API, and where the 30 to 50 percent savings come from.

​A practical look at how tone, body language, and simple answer structures make you memorable in interviews, even when other candidates have more experience.

A look at how seven pivots across two years led to Aside, and why overlooked problems, fast feedback cycles, and conviction matter more than a perfect idea.

We'll explore how founder-led content can build a brand people remember, turning demos, failures, and daily updates into users, leads, and investor interest.

​In this session, we'll move beyond the theory and look at the hidden force shaping our industry. We will explore how to transition from fragile architectures (where one error causes a cascade) to systems that leverage randomness to improve.

We’ll explore how AI-driven impact analysis can help identify what actually needs to be tested, reducing noise and focusing on real risks.

In this presentation, we will explores how to model and handle errors in Kotlin using a domain-driven approach.

I'll share how I completely rebuilt an LLM app from scratch in 10 months as the sole designer at a fast moving AI startup.

I will be focusing on one of the most common sources of friction between designers and developers: the gap between what is designed in Figma and what actually gets implemented.

I will show how Tilt helps you run a real microservices stack on a local Kubernetes cluster while keeping the same Kubernetes resources and configuration model you use in production.

We’ll explore how to effectively use a service mesh and GitOps-style workflows to make multicluster secure, reliable, and observable, the pros and cons of multi-cloud rather than simple multicluster, and how new technologies like federated Services affect the multicluster world.

Learn why and how one developer migrated their popular Korean baseball cheer song streaming app, from Firebase to Supabase, focusing on cost, queries, and developer experience.

Discover a practical guide to migrating a PostgreSQL database from Render to Supabase, including overcoming challenges with free-tier restrictions and version compatibility.

Screenshow is a powerful screen recording tool that generates professional, beautiful, and engaging video demos complete with motion graphics and 3D effects. Built with Supabase.

The speaker shares his experience and tips for rapidly building and launching a Minimum Viable Product (MVP) using Supabase, focusing on essential features, scaling, and security.

Insights from operating a Korean Large Language Model (LLM) leaderboard over one year and AI model benchmarking.

Learn about building a continuous performance improvement system for ClickHouse that leverages system tables, ties all metrics to schema versions, and uses a fast feedback loop with local benchmarks.

Learn about the latest features and acquisitions in ClickHouse, including the new parallel hash join and automatic join reordering, and see how they improve query performance.

Learn how the founder of TOPIK Easy6 uses PostHog to integrate analytics into their React Native app, prioritize features based on user behavior, and run experiments to understand user needs.

Explore how developer-first product analytics, specifically with the open-source tool PostHog, can provide crucial insights into user behavior, leading to better-informed product decisions.

Learn how to grow a mobile app business by examining the four stages of the user funnel: top of the funnel, app store page, onboarding, and the paywall itself.

Explore the pain points of a real-life project that required 933 potential code changes across six systems to introduce a new user class and how different access control models could provide a better way to handle permissions.

Follow a software engineer's journey of building a knowledge management and social app MVP for foreigners using Supabase and migrating from SvelteKit to React.

Hear the end-to-end story of how a small pain point with Kakao login on Supabase turned into an open-source contribution for a platform-optimized flow.