Bosch Power Tools
Bosch Power Tools in Mikulov is a service center for Bosch tools covering Germany, Czechia, Slovakia, Austria, and Hungary.
How the role unfolded
June — logistics intake
Worked on the receiving belt in the logistics department: unpacking shipments, learning day-to-day operations, and building working relationships on the floor.
July — office projects
Spent the full month in the office on two main projects — a company tour video and an internal employee/asset management app — plus two smaller initiatives: e-waste processing analysis and simplifying scoring for repaired machines.
Projects and outcomes
Four pieces of work — from a bilingual company film to a zero-cost internal app.
Company tour video
My role
- Script preparation — mapped how every department works before writing the narrative
- Shot dozens of high-quality video clips on site
- Coordinated an external drone specialist for indoor and outdoor aerial shots
- Wrote bilingual scripts (Czech and English)
- Produced natural voiceovers in ElevenLabs for both language versions
- Edited both videos in DaVinci Resolve
Results
Finished Czech and English tour videos (links to be added once published).
Internal management app
Problem
The site paid hundreds of thousands of CZK per year for a third-party program.
Director’s expectation
- Replace the tool with a cheaper solution
- Assumed I would build a complex Excel workbook
What I shipped
- A full internal web application
- Familiar UI so staff did not need to relearn workflows
- Faster than the previous tool
- Running cost: effectively zero
Technical approach
- Mapped the existing app’s behavior in detail
- Exported and migrated the databases
- Designed role-based access for different departments
- Used AI to analyze relationships between databases
- Built backend and frontend with AI assistance in Cursor
Tech stack
| Language | Purpose |
|---|---|
| TypeScript | Frontend SPA — React UI, routing, auth, API client |
| TypeScript | Backend — NestJS REST API, auth/RBAC, domain services |
| SQL | PostgreSQL schema, migrations, seeds |
| Python | One-off ETL: Excel → PostgreSQL |
| CSS / HTML | Styling and Vite shell / entry page |
| YAML | Local PostgreSQL and related infrastructure |
| Mermaid / Markdown | ER diagrams, specs, roadmap, design docs |
Result
Shipped a working application and cut annual software costs by hundreds of thousands of CZK.
E-waste processing analysis
Leadership assumption
There was a lot of e-waste that might generate more revenue than today.
Problem
The site produces e-waste, plastics, and paper. Some materials earn money; others do not.
Task
Find out whether — and how much — extra yearly revenue was realistically available.
Approach
- Studied which materials are produced and in what annual volumes
- Mapped e-waste processing stages and legislative requirements
- Compared buy-back price lists from several companies
- Looked for cheaper haulers for paper and plastics
Result
Recommended keeping the current setup: in-house e-waste processing would need heavy investment and operating cost and would not pay off; cheaper haulers existed, but the price gap was too small to justify switching.
Simplified scoring for repaired machines
Context
Every machine has a point score tied to employee bonuses. The Mikulov branch has repaired a very high volume of machines over time — the historical database holds more than 10,000 records.
Problem
Nearly identical repairs often had different point values just because series or model names differed.
Solution
- Used AI to group machines into main categories
- Created subgroups with current point values assigned
- Prepared columns for further nesting if needed later
- Handed the result to the workshop manager to finish with mechanics
Tools
- Excel — pivot tables, filtering
- Copilot — data processing
Skills and tools
- TypeScript
- React
- NestJS
- PostgreSQL
- Python
- Cursor
- DaVinci Resolve
- ElevenLabs
- Excel
- Operations
What I learned
Start on the floor to understand the real workflow, then use that context to ship tools and decisions leadership can trust — including recommending no change when the numbers do not support it.