Meet
From fixing Ferraris to building frameworks
13+ years turning problems into code that actually works
Chat with My AI Assistant
💬Try asking:
Privacy Notice
Your conversations are stored locally in your browser only and never sent to me. AI responses are powered by OpenRouter, which routes requests to various AI providers. While I cannot see your messages, the AI providers may retain your conversations according to their privacy policies, particularly when using free models.
A bit about me
With 14+ years of experience, I'm a developer, entrepreneur, and framework builder who went from diagnosing Ferraris to building AI-native platforms - and I still fix my own car.
It started with a car mechatronics apprenticeship at a Ferrari, Maserati, Aston Martin and Bentley specialist in Salzburg. I was the person who fixed the cars, handled the IT, and learned that every complex system - mechanical or digital - is just a set of interacting components waiting to be understood. That mindset never left. I co-founded my first company straight out of that apprenticeship, fixing phones and laptops, and never really stopped building things.
I led teams at Samsung Switzerland, founded and sold two IT companies, built trading bots with neural networks, wrote 20+ cross-platform plugins from a single codebase at Sovendus, and spent every spare hour on side projects that kept teaching me the same lesson: define things once and make them work everywhere. That lesson eventually became next-vibe. The widget engine, the trading platform, the email SaaS - they were all practice runs for the same core idea.
These days I'm building next-vibe and unbottled.ai full time. I've also shifted into a role I didn't expect to enjoy: working with AI as a genuine collaborator. I architect, review, and direct - AI implements at a pace I couldn't match alone. I miss writing every line myself sometimes, but shipping things that would've taken a team of five is a decent trade. I'm open to the right role or collaboration, but the main mission right now is finishing something that matters.
Professional Experience
My journey from car technician to framework builder - solving problems, leading teams, and shipping software across industries.
a42 / next-vibe / unbottled.ai
•Remote(Full-time)
Full-time on next-vibe and unbottled.ai after leaving Sovendus. My role has shifted into something I didn't expect to enjoy: architect, technical director, and code reviewer - with AI as the implementation partner. I spec the architecture, set the patterns, review everything, and direct AI agents to execute. We ship at a pace that would've required a team of five. I miss writing every line myself, but the output is hard to argue with.
Key Achievements
- Built unbottled.ai: free speech AI platform with 50+ models (mainstream, open, uncensored) - user-controlled filtering, privacy-first
- next-vibe: one definition.ts → web UI, CLI, MCP server tool, native app screen, AI-callable function - zero duplication
- Vibe Sense: TradingView-style rich timeseries analytics for any business domain - pipe endpoints together or describe a graph in plain language
- Switched to ts-go for 2M+ line codebase - check times from minutes to seconds, revealed architectural debt, drove systematic untangling
- Human-AI symbiosis: I architect and review, Hermes and Claude Code implement - patterns strict enough that both human and AI review are fast
- TanStack Start + Vite (dev) + Next.js App Router (prod) from the same codebase - zero DX compromise
- Hermes (local) and Thea (cloud) AI agents built into the platform - they monitor, delegate, and call endpoints as tools
Technologies & Tools
Sovendus
•Remote(Full-time)
Led the integration team. Partners were struggling with onboarding, so I built tools that let them help themselves. Turned a reactive support nightmare into a proactive developer experience. Nights and weekends during this period went into building what became next-vibe.
Key Achievements
- Built developer-hub.sovendus.com - documentation platform where partners actually find answers
- Created a testing app that lets partners diagnose 90% of their own integration issues without waiting for support
- Wrote 20+ platform-specific plugins (Shopify, Magento, Shopware, Flutter, React Native, Kotlin, Swift) - same core logic, different wrappers
- Designed unified codebase with platform-specific overrides - write once, deploy everywhere
- Shifted team mindset from 'answer tickets faster' to 'build tools so they don't need tickets'
- Left behind documentation and architecture that my successor can actually maintain
- Support tickets dropped dramatically because partners could fix things themselves
Technologies & Tools
a42
•Remote(Freelance / Side Projects)
Client work via Upwork alongside day jobs, plus building my own projects in every spare hour. Trading bots, SaaS tools, social media platforms - if it was interesting or someone would pay me, I built it. The side project work eventually evolved into next-vibe after learnings from Octane, anywhere-widgets, leadThing, and a dozen TypeScript full-stack experiments.
Key Achievements
- Built Octane - open source trading bot with visual strategy designer (still running)
- Built leadThing.dev - SaaS connecting 20+ newsletter platforms through one API
- Built anywhere-widgets - universal widget engine running on 20+ platforms from one codebase
- Created trading algorithms with ML-powered decision-making (actually 2x'd a test investment)
- Built hedge fund website connected to Octane for automated trading
- Built social media management and lead generation platform
- All of these eventually merged their best ideas into next-vibe
Technologies & Tools
Samsung Switzerland
•Switzerland(Full-time)
Led the Swiss business tech support team. Coordinated between subcontractors and global teams to solve enterprise hardware and software issues that nobody else wanted to touch.
Key Achievements
- Managed support team across multiple subcontractors - herding cats, but made it work
- Tackled complex enterprise issues that got escalated from everywhere else
- Worked with global teams to figure out problems nobody had seen before
- Built internal tools because the existing ones weren't cutting it
- Improved response times by actually fixing processes instead of just working harder
Technologies & Tools
netsepp KG (now viybs.com)
•Austria, Hallein(Full-time)
Co-founded an IT support company. Started with private customers fixing phones and laptops, grew into a full IT consultancy. Built the team, the systems, and eventually sold it to a former colleague.
Key Achievements
- Built the company from zero - found customers, hired people, set up processes
- Led a 5-person tech support team (who were way smarter than me, but I learned from them)
- Built the website and set up Odoo as our ERP and POS system because I'm too cheap to pay someone else
- Sold the company in 2019 to a former co-worker - it's still running today
Technologies & Tools
expert-service.at (now notebook-repair-corner.at)
•Austria, Salzburg(Full-time)
My first company. Fixed phones, laptops, tablets, printers - whatever broke, we fixed it. Started right after my car mechanic apprenticeship because I wanted to work with tech instead of oil changes.
Key Achievements
- Built the company from nothing - first customer, first hire, first everything
- Grew to a 5-person team (again, way smarter than me - I just asked good questions)
- Built the website and set up Odoo for ERP and POS with the team because paying someone felt wrong
- Sold to a bigger player in 2016 - they're still running it as notebook-repair-corner.at
Technologies & Tools
Formal Education
From fixing Ferraris to building software - the unconventional path
Salzburg, Austria
Leadership training and discipline (mostly just discipline)
Karner & Grossegger, Salzburg (Ferrari, Maserati, Aston Martin & Bentley specialist)
Diagnosing and fixing luxury cars, electronic systems, and learning that everything is just systems talking to each other
- •Passed final apprenticeship exam (officially a car mechanic)
- •Specialized in luxury vehicle diagnostics and retrofitting electronics
- •Became the unofficial IT support guy because I was the only one who understood computers
- •Attended seminars on personality, appearance, and communication (learned to talk to rich people)
Neumarkt (Department of Electrical Engineering)
Electrical engineering basics - where I learned Ohm's law and soldering
Things I've built & shipped
SaaS platforms, trading bots, custom frameworks, and developer tools. Some are live in production, some are still in development, and a few are retired but taught me valuable lessons.
Everything I've built for the past decade was practice for this. next-vibe is a SaaS framework built on a single principle: define it once, run it everywhere. One definition.ts file - with Zod schemas, field widgets, and examples - automatically becomes a web UI, CLI command, MCP server tool, native app screen, and AI-callable function. No glue code. No duplication. unbottled.ai is the flagship product built on it: a free speech AI platform with 50+ models where users control their own filtering level. Privacy-first, open source, and built for people who don't think someone else should decide what they're allowed to ask. The architecture absorbed everything - anywhere-widgets' SSR engine, Octane's data platform, leadThing's email layer, and years of building the same abstraction in five different ways before finally getting it right.
- •One definition.ts becomes web UI, CLI, MCP server tool, native app screen, and AI-callable function - zero duplication, detected at runtime
- •unbottled.ai: 50+ models (mainstream, open, uncensored) - users control their own filtering, not the platform
+5 more achievements...
Key Achievements
- •One definition.ts becomes web UI, CLI, MCP server tool, native app screen, and AI-callable function - zero duplication, detected at runtime
- •unbottled.ai: 50+ models (mainstream, open, uncensored) - users control their own filtering, not the platform
- •TanStack Start + Vite in dev, Next.js App Router in prod - same codebase, no compromise on DX or performance
- •Hermes (local AI) and Thea (cloud AI) built natively into the platform - they monitor, delegate, and call endpoints as tools
- •Vibe Sense: TradingView-style rich timeseries analytics for any business domain - pipe endpoints together or describe a graph in plain language and get interactive charts with events, signals, and drill-downs
- •Human-AI symbiosis in practice: I architect and review, AI implements - the framework's strict patterns make both faster
- •Type-safe end to end: schema is the single source of truth for validation, UI rendering, CLI args, AI tool descriptions, and docs
Code quality is an architectural decision, not a lint config. @next-vibe/checker wraps TypeScript, oxlint, and ESLint into one fast unified check with a clean summary - but that's just the surface. The real story is what switching to ts-go for a 2M+ line codebase revealed: type-checking that took minutes dropped to seconds, which opened a window into exactly why the code was slow to check. That visibility drove a systematic untangling of the whole codebase - circular dependencies eliminated, patterns made consistent, every module following the exact same structure. The result isn't just faster checks, it's code where human and AI review are both dramatically faster because everything looks the same. No `any`, no `unknown`, no `as X`, no `object`. If it compiles, it probably works.
- •Switched to ts-go for 2M+ line codebase - check times dropped from minutes to seconds
- •Performance visibility revealed circular deps and inconsistent patterns that were invisible before
+4 more achievements...
Key Achievements
- •Switched to ts-go for 2M+ line codebase - check times dropped from minutes to seconds
- •Performance visibility revealed circular deps and inconsistent patterns that were invisible before
- •Systematic untangling: every module now follows the exact same structure - human and AI review are both dramatically faster
- •~200 type assertions down from baseline and counting - targeting <10 core generic utils where casting is genuinely unavoidable
- •Bans: no any, no unknown, no object, no as X, no as whatever - zero exceptions, zero excuses
- •Ships as @next-vibe/checker on npm - drop-in quality gate for any TypeScript project
Think TradingView, but for your own company's data. Vibe Sense brings professional-grade timeseries analytics - the kind that pro day traders use to spot signals in market noise - to any business domain. Connect your next-vibe endpoints as data sources, pipe them together drag-and-drop, or just describe the graph you want and get rich interactive charts with events, signals, moving averages, anomaly markers, and drill-downs. The ideas came from years of building trading strategy platforms with Octane: if you can visualise a market with that level of richness, you can visualise anything - revenue, user growth, system health, whatever your domain looks like over time.
- •TradingView-style rich interactive charts - candlesticks, overlays, events, signals - for any business data source
- •Pipe next-vibe endpoints together visually or describe a graph in plain language and get it built
+4 more achievements...
Key Achievements
- •TradingView-style rich interactive charts - candlesticks, overlays, events, signals - for any business data source
- •Pipe next-vibe endpoints together visually or describe a graph in plain language and get it built
- •Technical analysis toolkit applied to business metrics: moving averages, anomaly detection, trend signals
- •Any timeseries domain: revenue, user behaviour, infrastructure health, content performance, trading
- •Evolved from years of building Octane's real-time market data and indicator engine
- •Integrated into next-vibe - every endpoint can become a Vibe Sense data source automatically
Open source trading bot platform with a visual strategy designer that actually works. Drag-and-drop block-based strategy building with 60+ indicator blocks, multi-threaded backtesting across all CPU cores, neural network integration for market prediction, and live execution via 50+ exchanges through CCXT. Still running. The real contribution wasn't the bot - it was building a platform where complex real-time data, technical analysis indicators, and visual pipelines all composed cleanly together. That architecture became the foundation for Vibe Sense.
- •Visual Strategy Designer: drag-and-drop pipeline with 60+ indicator blocks and 15+ evaluator blocks
- •Multi-threaded backtesting - parallel execution across all CPU cores for fast iteration
+5 more achievements...
Key Achievements
- •Visual Strategy Designer: drag-and-drop pipeline with 60+ indicator blocks and 15+ evaluator blocks
- •Multi-threaded backtesting - parallel execution across all CPU cores for fast iteration
- •Neural network evaluator: CNN, LSTM, RNN, Transformer architectures trained on real market data
- •TensorFlow model trained on 100+ stocks and cryptos - 2x'd a live test investment before I trusted it enough to scale
- •Live execution via CCXT: 50+ exchanges, paper trading and real money modes
- •Genetic algorithm-based parameter optimization for strategy tuning
- •Real-time data pipeline and indicator composition architecture evolved into Vibe Sense
Built 20+ integration plugins from scratch - for a platform that had zero before I joined. One unified codebase, platform-specific behaviour as overrides on a shared core: Shopify, Magento, Shopware, WooCommerce, React, Vue, Svelte, Flutter, Swift, Kotlin, React Native, vanilla JS. The same write-once-adapt-everywhere principle I'd been developing in side projects, now applied at professional scale with a team. Each plugin that shipped freed up more support time, which we reinvested into better tooling, which freed up more support time. The whole Sovendus developer ecosystem was one self-reinforcing cycle.
- •20+ integration plugins built from scratch - none existed before
- •Unified architecture: shared core logic, platform-specific overrides - not 20 separate codebases
+4 more achievements...
Key Achievements
- •20+ integration plugins built from scratch - none existed before
- •Unified architecture: shared core logic, platform-specific overrides - not 20 separate codebases
- •Native mobile SDKs: Swift (iOS), Kotlin (Android), Flutter, React Native
- •Web framework plugins: React, Vue, Svelte, vanilla JS, plus PHP for Shopify/Magento/Shopware
- •Single source of truth - fix it once, it's fixed everywhere
- •Self-reinforcing quality cycle: better plugins → fewer support tickets → more time to improve
Browser extension that catches integration mistakes the moment they happen. Near 100% detection rate across all Chromium and Firefox browsers. 90% of cases pinpoint the exact issue with actionable detail; the other 10% produce a structured error with reproduction data so the team can debug in minutes instead of days. Before this existed, QA was three students doing manual testing with a week of back-and-forth between them and dev. After: one person, instant feedback, partners self-diagnosing their own issues. This was the final piece that completed the virtuous cycle.
- •Near 100% detection rate for integration issues across Chromium and Firefox
- •90% of cases: exact error with fix guidance. 10%: structured reproduction data for the team
+4 more achievements...
Key Achievements
- •Near 100% detection rate for integration issues across Chromium and Firefox
- •90% of cases: exact error with fix guidance. 10%: structured reproduction data for the team
- •QA headcount: 3 students doing manual testing → 1 person with automated coverage
- •Integration feedback loop: 1 week of back-and-forth → instant, in the developer's browser
- •Partners fix their own issues - support tickets dropped to near zero for integration problems
- •Completed the cycle: docs + plugins + testing tool, each making the others better
Documentation that stays in sync because it lives next to the code that owns it. Custom server-side markdown renderer that pulls docs directly from each plugin's repository at render time - no separate docs repo, no manual sync, no stale PDFs. Partners find answers themselves. This was the first domino: better docs meant fewer basic support questions, which freed up time to improve plugins, which reduced integration errors, which freed up more time. The whole Sovendus developer ecosystem improvement story started here.
- •Custom SSR markdown renderer pulling docs live from each plugin's own repository
- •Replaced scattered stale PDFs and versioned Word docs with one always-current platform
+3 more achievements...
Key Achievements
- •Custom SSR markdown renderer pulling docs live from each plugin's own repository
- •Replaced scattered stale PDFs and versioned Word docs with one always-current platform
- •Docs stay in sync by design - they live next to the code, maintained by the same people
- •Support tickets dropped by half within months - partners found answers without asking
- •First domino in the virtuous cycle: fewer tickets → more dev time → better plugins → even fewer tickets
