Bewerbung.AI
Bewerbung.AI — German Application Platform
Agent-driven Lebenslauf and Anschreiben, job-fit coaching, Lambda PDF export, and analytics — tuned to the German hiring market.
Timeline
2023 – 2024
Stack size
15 technologies
Role
Senior Engineer (Consulting)
Company
Bewerbung.AI
Overview
Overview
Bewerbung.AI is a German-market application platform. It generates a full résumé (Lebenslauf), agent-driven cover letters (Anschreiben), and runs a multi-stage AI coaching flow tuned to German hiring conventions — Lebenslauf structure, Anschreiben formality, and Zeugnisse handling.
- I built across the full stack: React + RSPack editor with V2 resume/cover-letter architecture (Redux Toolkit, RTK Query, dedicated cover-letter agents and hooks), Express/Mongo backend, and serverless PDF export on AWS Lambda so heavy document renders do not block the API.
- Product instrumentation uses Statsig, Mixpanel, and Sentry with deferred third-party loading so analytics does not dominate cold start. AI usage is metered through Stripe credits with eval-driven prompt versioning to keep coaching quality stable as models and traffic scale.
Problem
Problem
German Bewerbungen follow specific conventions and expectations. Generic résumé builders miss format, tone, and role fit. Candidates need Lebenslauf and Anschreiben quality, a honest job-fit check against postings, and exports they can trust — without hiring expensive coaches or burning credits on low-value model calls.
Architecture
Architecture
Frontend
Backend
AI & agents
Export & infra
Product & analytics
Features
Features
- AI Lebenslauf editor with section-level rewriting, live preview, and clean PDF output across many templates.
- Agent-side Anschreiben (cover letter) generator — dedicated flows and V2 cover-letter state so agents personalise tone, structure, and role-specific arguments for German formality.
- Job-fit check: coaching agents grade applications against role requirements with gap analysis, discovery, and interview-prep stages so candidates know what to fix before they apply.
- Template-safe document rendering — publication links, rich text, and structured fields stay consistent from editor → API → PDF (desktop and mobile forms).
- AWS Lambda PDF jobs offload heavy export work from the main API; layout preservation on save so round-trips do not drop sections.
- Proactive improvement emails — applications are scored continuously and users receive concrete next steps via Postmark.
- Stripe paywall, full-screen pricing modal, and a credits system that meters AI usage and protects unit economics.
- Analytics setup: Statsig experimentation, Mixpanel product events, and Sentry — loaded on idle/deferred paths and isolated in Rspack vendor chunks so instrumentation does not dominate main-thread work.
- Cost optimisation: credit-gated AI calls, eval-driven prompts, bundle tuning (Rspack experiments, deferred analytics), and PDF strategy that keeps large artifacts off hot API paths.
AI systems
AI systems
- Cover-letter agents that draft and refine Anschreiben with German-market tone, tied to resume context and the target role.
- Coaching agents for job-fit scoring, gap analysis, and section-level rewrite suggestions against posting requirements.
- Proactive email agent that scans application state and surfaces highest-impact improvements.
- Eval-driven prompt management and credit budgets so quality regressions and runaway inference spend are caught early.
Outcomes
Outcomes
- Made high-quality German Bewerbungen accessible without expensive coaches.
- Cut time-to-application from hours to minutes with agent-assisted Anschreiben and trustworthy PDF exports.
- Improved submission quality through job-fit checks, proactive coaching loops, and template-correct exports.
- Kept SaaS unit economics healthy via credits, deferred analytics, and Lambda-backed PDF generation.
Technologies
Technologies
FAQ
FAQ
How is Bewerbung.AI different from generic résumé builders?
It is localised for the German Bewerbung process — Lebenslauf structure, Anschreiben tone, Zeugnisse handling — with agent-driven cover letters, job-fit grading against postings, and proactive improvement emails.
How does the job-fit check work?
Coaching agents compare the candidate’s Lebenslauf and Anschreiben to the target role, run gap analysis, and surface concrete edits — discovery, grading, and interview prep — before the user applies.
Why AWS Lambda for PDFs?
PDF export is CPU- and memory-heavy across many templates. Lambda jobs isolate that work from the Express API so document renders scale without blocking saves or editor requests.
More work