Alex Beverage

Every translator outputs native-speaker text. For a learner, that's text you can't read, can't use, and can't learn from. Levelang closes the gap. It translates at your proficiency level, in written or spoken form, in casual or formal register, and with mode and mood overlays. One input, up to 16 meaningfully different outputs per language. No other tool does this. The same constraint-based approach works within a single language. Casual speech becomes formal writing, a learner's rough attempt becomes natural speech, and dense jargon becomes something a beginner can read. The system spans three surfaces: a FastAPI backend, a React Native app with voice I/O, and an MCP server that exposes level-aware translation to AI assistants as consumable infrastructure. The quality methodology is the competitive advantage. Prompts are adversarially hardened and validated by automated test suites with AI judges, producing 100% pass rates across every supported language. Built an agent skill system that runs attack → tune → validate cycles in minutes, turning prompt quality into a repeatable engineering discipline. Made every decision across the product: architecture, infrastructure, prompt engineering, product strategy, linguistic design, go-to-market positioning, and the pivot from a consumer app to a dual-surface platform (app plus translation-and-transformation-as-a-service API). Stack: Python, FastAPI, PostgresSQL, pgvector, RAG, LLMs, MCP, Redis, React Native, Expo, Astro.

Analyzed, investigated, and saved significant revenue on 3rd-party solution costs measured in the millions of euros. Built, upgraded, and optimized CI/CD systems as a service for many teams across all of tech, all of which were customizable for their specific needs as they liked. Results were from 5 teams to most of our tech organization within a few months. Refactored, and repaired, and fixed compliance requirements in advance of GDPR for our internal LDAP systems. Skills: Java · Spring Boot · Python (Programming Language) · Continuous Integration (CI) · Amazon Web Services (AWS) · Software as a Service (SaaS) · Jenkins · Databases (Postgres) · Debugging · Git · Infrastructure · Continuous Delivery (CD) · Computer Science · Docker Products · Kubernetes · REST APIs

+Built out what started as a UI-oriented test platform, and expanded it to be able to cover the entire spectrum of components of our product – 40+ services, several backend applications, 5 internal websites, and 2 customer facing websites. +Delivered test-data-as-a-service solutions to better allow developers and QA engineers to follow test driven design as they built their services by allowing them to enjoy the full richness of our test automation platform at a minimal entry cost. +Isolated services from major external dependencies with Spring dependency injection. +Mentored and trained a team of QA engineers to better write test automation, was able to deliver increases in test coverage week over week, and bugs found by automation release over release. +Cut release times from QA by significant margins (in some cases by as much as 90%+), and moved many of our deployments over to continuous deployment.