Modernising an ML-driven subsurface platform
Principal engineer on QEarth and QLog, the seismic, well-log and real-time drilling workflows that energy operators run on Quantico's machine-learning platform.
- Client
- Quantico Energy Solutions
- Period
- May – Nov 2026
- Sector
- Energy
- Role
- Principal Software Engineer, Architecture roadmap owner
Quantico’s platform applies machine learning to subsurface data: seismic volumes, well logs and live drilling feeds. I joined as principal engineer to modernise it.
The first deliverable was an architecture roadmap: one monorepo for sixteen repositories, a uniform dashboard, PostgreSQL consolidation, SSO and feature flags. I then carried out the central step, merging QEarth and QLog into one codebase with shared libraries, one React UI, one gateway and one database.
Alongside the restructuring I hardened multi-tenant security and shipped product features in FastAPI and React: content search, workflow groups and project import.
Agentic software design
The team works agent-first. I set up the repository context (CLAUDE.md), agent briefs and test gates, so that Claude Code and Codex implement changes and verify them against the test suite before a human reviews them.
Key challenges
- Sixteen repositories, two products and diverging UIs had to converge without stopping delivery to operators.
- Multi-tenant data where every file belongs to a team, and every file operation must respect that.
- Raising throughput with AI coding agents while keeping every change reviewable and verified.
Architecture highlights
- One monorepo for QEarth and QLog, with shared libraries and a shared React UI.
- A single gateway and one consolidated PostgreSQL database instead of per-product stores.
- SSO and feature flags as platform services, so features roll out per tenant.
- Team-scoped storage with ownership checks on every file operation and path-traversal protection.
Stack
- Python
- FastAPI
- React
- TypeScript
- PostgreSQL
- SSO
- Feature flags
- Claude Code
- OpenAI Codex
