What the work actually involves

You own the AI capabilities inside the UC-003 Billing Prep program: a classification model that decides whether a billing line item is reimbursable under contract, anomaly detection that catches charge spikes and data-quality problems before a human sees them, and in-month monitoring agents that surface incomplete accruals mid-cycle rather than at period close. On top of that sits a Gemini-based copilot for billing analysts — explain why something was flagged, suggest a classification, retrieve similar historical cases. Day to day you are in Vertex AI (Workbench, Pipelines, Model Registry, Feature Store, Prediction endpoints), writing Python, wrapping models in FastAPI or Flask, and negotiating service contracts with a backend engineer and feature/label pipelines with a data engineer.

The governance half of the job is not decoration. Every model output has to carry a confidence score and an explainability artefact, nothing may bypass the human review workbench, all inferences are logged with provenance for audit, and you are expected to run bias checks across vendor, cost centre and contract type and write model cards that survive a compliance review. If you have shipped ML where the only success metric was offline AUC, this role will feel different.

What the screen is looking for

Turing's process front-loads a structured interview before any client-facing round. Expect it to push on specifics: which Vertex AI components you personally configured versus read about, how you set thresholds on an imbalanced classifier, how you enforced structured output from Gemini and what you did when it drifted off schema, and whether you have worked in a financial or ERP-adjacent domain where a misclassification has money attached. Vague ownership claims ("we used Vertex AI") collapse quickly under follow-up; be ready to name the metric, the baseline and the delta.

Logistics

  • Location: Hyderabad, on the listing as stated — treat this as an onsite or hybrid seat unless the recruiter confirms otherwise. It is not an async remote role.
  • Engagement: full-time, reporting to a Solution Architect / Tech Lead, inside an agile delivery cadence with SME feedback loops and UAT support.
  • Experience band: 4–7 years of production ML/AI, despite the "Principal" framing in the board title.
  • Pay: undisclosed on the posting. Ask early; do not infer a band from the seniority label.