Munich, Bavaria, Germany
I used to lose hours daily to my phone because the apps are designed that way. I tried Screen Time, One Sec, Opal. Every single one either didn't work or used dark patterns to sell itself. So I built ScrollJail the AI that asks "why?" every time you open a distracting app on autopilot. You can't autopilot through a conversation. → scrolljail.com
Developing an AI accounting platform for enterprise finance teams.
• Developed and delivered the "Deploy Bedrock Chat: Your Enterprise AI Chatbot" workshop for AWS Generative AI Immersion Day, achieving a CSAT score of 4.7. • Owned the design and implementation of the deployment, bot creation, RAG, authentication, and security sections. • Added MCP feature to open source Bedrock Chat repository (Issue 866). • Completed AWS Tech U training program; consistently raising the bar in technical assessments and certifications.
We’re building mailflow, an AI assistant that connects to your inbox, drafts and automates email replies using public and internal data. It classifies emails, triggers workflows, and tracks performance—turning email into a measurable, automated channel.
• Enabled 150+ users to quickly create job-tailored resumes and cover letters by developing and deploying JobBot, an AI-powered assistant, which saves a user an average of 2 hours per application. • Deployed a secure and scalable Java (Spring Boot) monolith backend with a PostgreSQL database and a Vue.js frontend using Docker, GitHub Actions and Azure Container Registry, reducing operational costs to about 10C per month while maintaining 24/7 uptime without cloud start delays and a median response time of 3ms. • Integrated fine-tuned OpenAI model and Gemini API function calling for intelligent resume parsing, job description alignment, and personalized cover letter generation, facilitating real-time interactions through a WebSocket-based chat interface.* • Designed a user-friendly and bilingual (i18n ready) UI for JobBot, improving user experience and accessibility for English-speaking users, resulting in smoother and more efficient interactions.* • Implemented state of the art security measures including CSP, Double-Opt-In sign-up, session-based authentication, CORS, CSRF protection, and Google reCAPTCHA to enhance application security.
• Led development of a scalable cloud-based flight test data plotting application using Python, Flask, Plotly, React, and Kubernetes, processing 20+ GB files and visualizing GPU-rendered graphs from millions of data points. • Optimized large-scale data visualization, reducing dataset size and noise by 95% with a custom downsampling algorithm (second derivatives, decimation, Gaussian filter), resulting in 200% less memory usage. • Migrated the most used Palantir dashboard (100+ daily users), enhancing data accuracy and reducing errors in data processing by 90% through updating and optimizing with PySpark, and deprecating legacy data sources. • Automated CI/CD pipelines, saving 5 minutes per deployment and reducing overall deployment time by 2 hours per week, using Azure Container Registry, Docker, and GitHub Actions. • Led knowledge-sharing sessions for 10+ data professionals on cloud technologies, distributed systems, neural networks, and GPTs, enhancing team expertise and fostering growth.