Alex Stewart: keeping transcription errors out of the molybdenum certificate
Five stages in one workflow, from intake of the concentrate to certification. Every certificate can be followed back to the batch and the intake it came from.
- Sector
- Mining
- Role
- CTO
- My part
- Technical direction, development and project management.
- Period
- Latest release in 2026
- Status
- Operations in production. Document library in use and in development
- Team
- Seven to eight people over the life of the project
System blueprint
The flow, in text
- Intake of the molybdenum concentrate.
- The material is split into batches.
- Moisture is measured.
- Chemical analysis.
- Certification: every certificate can be followed back to the batch and the intake it came from.
- The system generates certificates and reports as PDF, Word and Excel. Underneath, the platform: NestJS, MongoDB and AWS.
Platform: NestJS · MongoDB · AWS Cognito and S3 · Diagram. Not live data.
Product in use
Problem and constraints
Before a certificate of analysis is issued, molybdenum concentrate goes through intake, is split into batches, has its moisture measured and is chemically analysed. Each step is done by a different person, and each leaves a figure the next one depends on.
If that chain lives in spreadsheets, a transcription error travels all the way to the certificate, which is the document a sale is settled on. The lab’s molybdenum operation is with Codelco.
Constraints
- A sale is settled on the certificate: a transcription error must not reach it.
- Each stage is done by a different person and leaves a figure the next one depends on.
Decisions and architecture
Everything runs in one workflow with full traceability, following the material: intake, batches, moisture, chemical analysis and certification are recorded in the same system, and every certificate can be traced back to its origin.
The system produces the documents. Certificates and reports are generated as PDF, Word and Excel from the recorded data, with no copying and pasting.
The lab’s procedures and documents live in a separate document library, called Biblioteca: a folder hierarchy, multi-level permissions and a rich-text editor (Editor.js).
Both pieces share a foundation: a NestJS API on MongoDB, authentication on AWS Cognito and files on S3. The web app is React with Vite.
Seven to eight people have worked on it over the life of the project. My part is the technical direction of the team and the architecture.
Results
The operations platform is in production and covers the full workflow, including a laboratory module.
The document library is in use and still being developed.
- 5stages in one workflow: intake, batches, moisture, chemical analysis and certification.
- 3output formats for certificates and reports: PDF, Word and Excel.
- 2systems: lab operations and a document library with tiered permissions.
Limits
- This case does not publish volumes or data from the lab’s operation.
- The document library is still in development.
Do you have a system that cannot be touched?
Tell me the problemWant the full career history?
View the CV