Selected Work
Five pieces of work from training, Swisscom rotations and a hackathon. They range from AI interfaces to platform engineering and automated processes. Keep scrolling →
Apprentice application developer at Swisscom; full-stack, AI/ML integration and MCP. Leads Team Halo with 33 apprentices.
Learn more
code is
craft.
01: About me
I build applications that use AI. Across three project rotations at Swisscom I have moved from low-code all the way to production-grade platform development. My focus: the Model Context Protocol (MCP), which lets AI assistants access and interact with enterprise platforms securely.
Based in Waltenschwil in the canton of Aargau. Two sisters, both in IT as well. I tinker on OpenClaw in my own time, an idea for configuring agents for companies. When I'm away from the screen: the gym after work or outdoors with friends and my girlfriend.
02: Selected work
01 / 04
Live translation · current project
Simultaneous Translation
Transcribes speech in real time and translates it on the fly. This is ideal for events where not everyone shares a language. A context field lets you supply names and technical terms up front, so the recognition gets them right.
Go · Next.js · Full-stack
MCP Gateway
Platform built at Swisscom: a Go backend bundling multiple MCP servers behind a single endpoint, plus a Next.js management UI with SSO and gRPC.
View live ↗ Swisscom corporate network only
Hackathon · AI & security
swissscam
Detects scam calls while they happen. Whisper transcribes the conversation locally, a fine-tuned MiniLM scores the text heard so far and warns before the demand arrives. Everything stays on the device. No recording leaves the machine. Built in 48 hours at the swiss{si}weeks in Zurich for Swisscom's identity-fraud challenge. My part: the machine learning and the frontend design.
GitHub repo ↗
AI · Full-stack
AI chatbot todo app
A todo app fully controlled through chat. It uses natural language instead of clicks, with a React frontend and Spring Boot backend deployed on AWS.
Machine learning
Football predictions
ML model for predicting match results, from data preparation to the trained model. It rates individual matches by probability and simulates the whole 2026 World Cup: the winner comes out as Spain. When the project was built the play-offs had not been decided yet. That is why Italy runs through the bracket even though it did not qualify.
GitHub repo ↗
03: Journey
2025 to 2027
Team Halo
Full-stack with a focus on AI integration and machine learning: MCP client/server implementations, CRUD apps with React & Spring Boot, AWS deployments, ML models. After a year on the team I took over leading Team Halo and now lead 33 apprentices.
2025
Power Builders
Low-code development with Microsoft Power Apps: Super League app, certificate management, workflow automation.
2024 to 2025
Apps Team
Interactive applications including a memory game and help resources for customer support. My entry into the craft of frontend development.
“Aim for the moon.
If you miss, you may hit a star.”
References: Nithursan Naguleswaran (LB) · Michel Haueter (host)