• Location: Würzburg, Bavaria
• Work authorisation: EU Blue Card (2025)
• Work mode: Remote / Hybrid
• Email: nick_suomi@icloud.com
• LinkedIn: https://www.linkedin.com/in/nikita-gerasimenko
• GitHub: https://github.com/NickSuomi

I think in systems and speak business. Give me an outcome; I’ll map the path, surface trade-offs, and ship. Formal foundations let me learn fast and go deep where it matters. Everything on my CV is production work that delivered value, including modern AI built as real systems, not demos. I’m deliberate about projects; if the product and team values resonate, I commit fully and stay with it until the result sticks.

Product: Intranet / knowledge & communications platform with real-time wiki/chat for organisations, file management, kanban boards, video calls, rich-text editing, OnlyOffice-based collaborative editing and AI Knowledge Base on RAG; shipped on web, Electron desktop, and iOS/Android (Capacitor).

Main Stack: Vue 3, Pinia, TypeScript, Vite, Electron (macOS/Windows), Ionic/Capacitor/Cordova (iOS/Android)

1) AI Knowledge Base (OpenWebUI + server‑side RAG)
Why - first paying client needed in‑product answers over org docs; keep one loop; meet ACL & DE‑hosted constraints.
What - in‑app Q&A with inline source snippets and deep‑links to doc/version.
How - backend indexer -> OpenWebUI collections per network/group; server‑side calls via env proxy; periodic re‑index; supports OnlyOffice/PDF/wiki; ports‑and‑adapters.
Result - reviews stayed in‑product; less context switching; shipped, adopted, and resold.

2) Desktop App (Electron) - cross‑platform shipping, signing & auto‑updates
Why - customer needed a native desktop channel and controlled rollout; stack was web‑only.
What - production Electron app for macOS/Windows with code signing and silent auto‑updates.
How - parametrised Bash CI -> artifacts -> update endpoint; Dockerised updater; hardware‑token signing; custom update manager (progress/logs); migrated PDF preview to OnlyOffice and load‑balanced the preview tier.
Result - desktop became first‑class; one‑command releases; preview stability recovered; built for a paying client and resold.

3) Observability & Incident Response (Sentry)
Why - monthly releases produced unpredictable regressions; triage depended on users.
What - single focal point for failures and feedback.
How - self‑hosted Sentry (env/release tags, session replay, perf, privacy/quota tuning) + feedback widget; Axios interceptors; typed error taxonomy.
Result - MTTR ~3 days -> ~12 hours; context‑less tickets eliminated; fewer mid‑sprint interrupts; better customer trust.

4) Rich‑text Wiki Versioning & Diff
Why - teams needed to see who changed what, compare versions, and safely restore.
What - version history + inline/side‑by‑side diff; deep‑links to a specific version.
How - Node.js diff service (temp git repo -> patch) + client render via diff2html; size guard for large pages; restore creates a new version (append‑only).
Result - faster reviews, fewer escalations, auditable history; built for a paying client and resold.

Product: 3D visualisation for a university campus on Autodesk Platform Services (APS, formerly Forge) Viewer with IoT sensor overlays (temperature, humidity, CO2, illuminance) on the BIM model.

Main Stack: Vue 2, Quasar, JavaScript, APS (Forge) Viewer, IoT API.

1) IoT Heatmap & Timeline inside the 3D viewer
Why - instant, repeatable 3D views (heatmaps, jumps to elements, filters) and time‑lapse playback without manual camera work.
What - control dashboard (toggle heatmaps/sensors, navigate to coords/elements, save/restore presets, timeline with interval/speed).
How - APS Viewer API (extensions/events, value->colour mapping), IoT API (live/historical), serialised camera/layer state; targeted legacy refactors.
Result - one‑click exact views and slideshow loops; analysis moved into the in‑viewer experience.

Product: Part of an archive digitisation suite (hardware scanners + embedded software + client web app) used by institutional/government customers to scan and organise large volumes of legacy documents. I worked on the client web application for archivists/operators: navigating deep hierarchical collections and previewing scans.

Main Stack: React, Node.js, Express.js, Postgres, Webpack.

1) Deep hierarchy navigation & high‑volume preview performance
Why - operators needed fast navigation of deep trees and quick previews; initial lists lagged at scale.
What - document browser with smooth traversal, fast pagination, and inline previews.
How - optimised list/tree rendering and update batching; progressive loading for large collections and previews.
Result - snappier day‑to‑day work on large hierarchies; quick jumps and previews without stalls.

Delivered assorted short‑term projects across e‑commerce, dashboards and site builds; handled migrations and integrations as needed.

• Frontend: Vue 3, Pinia, TypeScript, Vite, TinyMCE; Electron (macOS/Windows), Ionic/Capacitor/Cordova (iOS/Android), Sass, OnlyOffice, IndexedDB, WebSockets;
• Backend: Node.js, Express.js, SignalR, WebRTC, Twilio;
• AI: OpenAI API, OpenWebUI, RAG;
• CI/CD: Nginx, Bash, GitLab CI, Fastlane (iOS/Android);
• Tooling: Sentry, ESLint, Prettier, Jest/Vitest, Git hooks, Postman, Datadog;
• Infra & Cloud: Docker, Firebase;
• Architecture: Ports and Adapters; Function-Core/Imperative-Shell; SOLID, DRY; Feature-Sliced Design (FSD); monorepo conventions;
• Security: Enforced no client-side secrets, strong ACLs/permissions, XSS mitigation (DOMPurify, CSP), hardened headers, pentest-proven, and DE-only data residency;

• Frontend: React, Nuxt.js, Quasar/Vue 2, APS (Forge) Viewer, three.js;
• Backend: .NET/C#, Ruby, Bun, SQL Server, Postgres;
• AI: Mistral API, Qdrant Vector DB, OpenAI Embeddings API;
• CI/CD: GitHub Actions;
• Tooling: Storybook, Selenium;
• Architecture: Domain-Driven Design, monorepo conventions, microfrontends;

B.Sc., Applied Informatics - Saint-Petersburg State University of Industrial Technologies and Design (2019–2023)
B.Sc., Chemistry - University of Strasbourg (2016–2019)

• English: C1; French: C1; German: B1; Russian: Native; Japanese: A1

• Engineering preferences: FP‑leaning in TypeScript; Effect‑TS; neverthrow; ts‑pattern; architecture - Ports & Adapters; Functional Core / Imperative Shell; FSD;
• Open source / community: Effect‑TS (active user)
• Interests: Foreign languages, tennis, cycling