Sergi Jajanidze

Split a web-component SPA serving two product lines into a shell and two independently deployed remotes, sharing exactly one copy of the framework runtime. Federation emits native ES modules, so remotes load through the browser's own `import()`, and routes moved to their owning remote one content type at a time.

Role

Lead engineer — architecture owner

Scope

A shell host plus two product remotes, each deployed on its own cadence · one import map shared by all three builds

Outcome

3 independently deployed apps across 2 teams, with release conflicts between them eliminated

Moved the design system and framework runtime out of three application bundles and onto a CDN, resolved by the browser from one import map. Webpack externals are derived from that map, so an unresolvable import can't ship, and a design-system upgrade is a version bump in one JSON file rather than a rebuild of three apps.

Role

Lead engineer — designed and built the dependency-resolution layer

Scope

One import map — design-system packages, framework runtime and federation entry points — shared by three independently deployed applications

Outcome

About 60% smaller application bundles

Decoupled search visibility from the client-side module graph: an Express middleware detects bots, renders the page in pooled headless Chromium, strips the scripts and returns finished HTML from the same origin and deploy that serves humans. The key decision is forcing the shadow-DOM polyfill before the app loads — otherwise a site built from custom elements serializes to empty tags.

Role

Lead engineer — design, implementation, rollout across sites

Scope

One versioned middleware package used by five public sites · configurable crawler list, two cache backends, per-site render settings

Outcome

Improved search indexing across 5 public sites

Built bilingual full-text search for three public sites as one Node service with a hand-built character trie in memory — no search cluster and no network hop per query. A prefix trie is the crude stemmer an agglutinative language like Georgian needs, and query-time transliteration lets Georgian typed phonetically in Latin find the same pages.

Role

Lead engineer — designed the service, wrote the first commit, owned it in production

Scope

Three content channels × two languages — one in-memory index per pair, rebuilt daily, with no search cluster to operate

Added an LLM query-understanding layer in front of an existing keyword index instead of replacing it, so questions and typos resolve to indexed terms — plus a streamed answer card grounded only in pages the index already returned. Every model call falls back to the raw query, so a provider outage degrades to the original search, never to an error.

Role

Lead engineer — sole implementer, backend and frontend

Scope

Two endpoints, one shared search component with two consumers, two locales

Outcome

50% more clicks on search results

Migrated a daily-use editorial CMS off a dead framework, running old and new apps side by side behind one reverse proxy so editors never saw a freeze. The real work was rewriting a form engine that generates the whole editing UI at runtime from user-defined content types — including multi-locale editing that Polymer had solved with DOM traversal.

Role

Lead engineer — engine owner

Scope

TypeScript rewrite of a Polymer 3 app · schema-driven form engine · one deployment serving several sites with their own locale sets

Outcome

Adopted across 5 product teams