Live: Internal / commercial platform · Code: Proprietary · Skills proven: Go (Golang), Python, PostgreSQL, System design, Microservices, ETL / ELT, Data modeling, Observability
The story
Context. Enterprise operations teams often run critical processes across disconnected tools with inconsistent documentation and manual handoffs. Worktrace Pro provides a platform for discovering, documenting, and automating operational workflows from real operational data.
(Note: This summary covers public-safe architectural responsibilities without sharing proprietary client names, internal schemas, or confidential business metrics.)
My role and constraints. As Senior Software Engineer at Worktrace (since Jan 2025), I architected the core workflow discovery and documentation platform, built automated solutions for complex operational tasks, and developed scalable backend systems and data pipelines to process large volumes of operational data.
Key decisions.
- Decoupled ingestion pipelines and workflow modeling: Separated operational event collection and ETL processing from the workflow discovery query layer so heavy data ingestion bursts do not degrade platform responsiveness.
- Relational modeling in PostgreSQL with service-level observability: Modeled workflow steps, ownership, and execution states cleanly in PostgreSQL while instrumenting backend services so pipeline bottlenecks and failure states are visible early.
What went wrong. Processing large volumes of varied operational data exposed edge cases where malformed or bursty records caused pipeline backpressure. Introducing stricter schema validation at pipeline boundaries and bounded processing stages kept ingestion predictable.
Result. Delivered the core platform architecture, operational automation workflows, and scalable backend data pipelines supporting enterprise workflow discovery and documentation.
The evidence
- Production architecture and backend data pipeline ownership in Senior Software Engineer role at Worktrace (Jan 2025 — Present)
- Clean separation of public architectural patterns from confidential employer/client data
- What I'd do next: extract non-proprietary pipeline pattern notes and Go/PostgreSQL performance learnings into public blog posts on hersi.dev