Niklas Menzel‑Sandic

German leisure airline. Greenfield booking platform: revenue-critical surfaces including an embedded React/Vite search widget and the shopping cart. Worked in Team Search/Cart, one of twelve cross-functional teams (PM, BA, engineers and designers) on the platform, and the only one covering two product areas.

• Product & engineering ownership — Owned major product and engineering workstreams across the greenfield booking platform, shaping feature scope and technical direction with product and design, and taking work through architecture, implementation, release and production iteration with the backend and platform teams.
• Performance — Improved LCP from ~4.2s to under 2.5s (Lighthouse) and reduced bundle size by ~28% through Vite/Rollup tuning, dependency audits, tree-shaking, dynamic imports for non-critical modules, critical CSS extraction and render-path prioritisation.
• Flight Search Widget & Shopping Cart — Led the technical design — component boundaries, state-management strategy, auth-dependent routing, widget-host integration and API contracts with the backend teams — so later features had somewhere safe to land.
• Authentication — Owned the frontend OAuth/JWT authentication integration for the Search Widget, including protected-route behaviour, client-side token handling and defensive validation in user-facing flows, while aligning the contract with the API team.
• Notification system — Designed and built a global notification system for search and booking, modelling loading, success and error states consistently so failures in critical booking situations are clearer for users and easier to diagnose.
• Accessibility — Drove improvements across key journeys through semantic HTML, ARIA corrections, focus management, keyboard navigation and recurring axe-core checks; documented repeatable patterns for sustainable remediation.
• Observability — Established production monitoring for search and booking flows with Splunk and Sentry, so production issues could be identified, prioritised and fed back into engineering work faster.
• Enablement — Authored the first official updated developer-onboarding documentation, drove knowledge sharing across an engineering Community of Practice of ~32 developers including presentations on secure-coding standards, and owned engineer onboarding — taking two engineers to a first production contribution through structured pairing, code review and documentation.
• Deployment & cloud — Handled application-side deployments, environment configuration and production debugging for AWS-hosted services and Redis-backed caching, in collaboration with DevOps and infrastructure teams.

~20-person startup behind a geospatial MLOps forecasting platform, used for wildfire and flood-risk analysis by German emergency services.

• Primary engineer — Worked inside a small product and engineering group and became the primary — and at times sole — software engineer responsible for the platform's software engineering.
• End-to-end ownership — Owned product and engineering work from customer feedback and early ideas through scoping, technical design, implementation, deployment and production iteration, shipping improvements on a daily or weekly cadence.
• Working with the founder — Worked directly with the founder/CEO and with data-science colleagues, using customer feedback as the primary input for product iteration while independently owning software-engineering decisions and turning data-science prototypes into production-facing interfaces.
• Services & pipelines — Developed backend services and data pipelines with Node.js, Docker, Kubernetes and GCP, with Firebase as the primary data store.
• Real-time performance — Diagnosed performance and scalability bottlenecks in a real-time fire-hazard-detection microservice across model logic, data processing and UI behaviour, and drove the improvements with the MLOps team.
• CI/CD & error tracking — Introduced GitHub Actions workflows for automated build, test and deployment checks, and operationalised Sentry-based error tracking for backend services.

Hands-on software engineering and consulting delivery for enterprise banking clients, entered with prior freelance production experience.

• Enterprise banking features — Independently implemented React and Java/Spring Boot features for major German banks including Deutsche Bank, ING-DiBa, Commerzbank and DKB, inside their own regulated codebases and senior-reviewed enterprise delivery processes.
• Feature responsibility — Took day-to-day engineering responsibility for assigned features from technical analysis and requirements clarification through implementation, testing, documentation and handover; senior consultants retained final architectural and client sign-off.
• Compliance-sensitive delivery — Worked directly in established production codebases and customer-specific technology stacks, investigating issues, proposing technical solutions and implementing changes under the reliability, traceability and secure-development requirements of regulated banking.
• Client communication — Collaborated with senior consultants and customer teams on technical requirements, reviews and implementation handovers across compliance-sensitive projects.

Independent client projects for SMEs in e-commerce, logistics and digital health, alongside university.

• End-to-end delivery — Owned end-to-end delivery of web applications for SME clients using React, Next.js, Vue, Node.js and TypeScript.
• Performance — Raised Lighthouse performance scores in selected client projects from ~60 to 90+ through caching, asset optimisation and component-level refactoring.
• Selected client work — Rentbox24, a full-stack storage-booking platform with REST APIs, PostgreSQL and payment integrations. Wellmo, a Vue.js questionnaire extension for a configurable digital-health platform, including UI/UX collaboration and automated testing.
• Delivery pipelines — Built GitHub Actions pipelines to automate builds, linting and test execution, making delivery more reliable.

An answer to one question: how do you adopt AI tooling in software delivery without weakening engineering accountability?

• Shipwright AI Framework — A Next.js companion repository with a responsible-AI-usage policy, review checklists, prompt templates, a human-in-the-loop model and an example delivery workflow.
• Shipwright MCP Server — A read-only TypeScript MCP server exposing those engineering standards to AI coding assistants, using lexical keyword search, Zod-validated inputs, path-traversal protection and unit tests for the security-critical paths.
• Role-specific workflows — Documented AI-assisted planning, PR review, security review, accessibility review and architecture review. Planned extensions: vector retrieval, GitHub PR metadata and richer editor/CLI integrations.