About the Role
The CTO Intelligence Hub is OPSWAT’s internal AI-powered data platform; dashboards and APIs that give leadership real-time visibility into AI adoption, financial performance, customer health, and sales. The platform runs on a fleet of Flask, FastAPI, and Streamlit applications backed by a Fabric → Parquet → S3 data pipeline, all tied together by an agentic coding system where GitHub Issues become production pull requests in under three minutes.
As AI Platform Architect, you own the application layer. You design and build new intelligence products, evolve existing dashboards, and work with the AI coding pipeline — writing the specs, reviewing the plans, and guiding the implementation. You are the “human senior engineer” in a team where your main coding collaborator is Claude.
What You Will Own
- Application architecture: Flask APIs, FastAPI microservices, Streamlit dashboards; schema and data-flow design across 12+ applications
- Agentic coding pipeline: creating well-scoped GitHub Issues, reviewing AI-generated implementation plans, approving or refining, merging PRs; the pipeline (claude-coder) takes issues from triage to production PR in ~3 minutes for routine work
- Data pipeline & analytics: extending the Fabric → Parquet → S3 pipeline; designing new analytics (percentile ranking, quadrant analysis, trend computation); HR enrichment and data quality
- Test-driven development: writing pytest spec suites that the AI implementer must satisfy; test files are the spec and cannot be modified by the AI
- Claude API / CLI integration: building features that call Claude for inference (content generation, analysis, RFP answering); working with OAuth-based CLI inference and direct API patterns
- UX & visualisation: Plotly dashboards, Jinja2 templates, React/Vite prototypes; maintaining the OPSWAT brand system across all apps
- Architecture documentation: keeping Confluence up to date as the system evolves; writing CLAUDE.md context files that teach the AI pipeline how each repo works
- New product design: identifying gaps in current intelligence coverage and proposing, scoping, and shipping new internal tools
You Will Be a Strong Fit If You Have
- 3–7 years of Python backend development (Flask, FastAPI, or equivalent)
- Strong data skills: pandas, SQL, Parquet, data modelling; you are comfortable designing a metrics pipeline from raw HR and usage data to ranked leader-board
- Experience working with LLM APIs (OpenAI, Anthropic, or equivalent) in production
- A test-first mindset: you write tests before code and believe the test suite is the specification
- Clear written communication because your primary “pairing partner” reads GitHub Issues, your ability to write precise, unambiguous requirements directly determines output quality
- Comfort reading and reviewing AI-generated code critically; you know when to approve, when to push back, and when to take the keyboard yourself
- Experience with REST API design and documentation
Nice to Have
- Streamlit or Plotly dashboard development
- React / TypeScript (for UX prototypes)
- Microsoft Fabric, Azure Data Warehouse, or similar cloud data platforms
- Familiarity with GitHub Actions and CI/CD pipeline design
- Experience building multi-language document processing (the RFP automation covers 14 languages)
- Prompt engineering and system-prompt design for agentic workflows
- OPSWAT product knowledge (MetaDefender, OESIS, CDR - not required but helpful)