Databricks AI Sales Platform Build
Publicada el 2026-07-19
Descripción de la oferta
**Job Title:** Databricks Engineer Needed — End-to-End AI Sales Analytics Platform (Genie, Unity Catalog, MLflow) --- **Job Description:** We need an experienced Databricks engineer to take a complete technical specification (attached) and implement it end-to-end — from raw data ingestion through to a fully deployed, tested, and monitored production application. This is not a partial build or a prototype: we're looking for someone to execute the entire attached implementation plan, section by section, exactly as specified, and deliver a working system. **About the project:** We're building a natural-language Q&A tool for our sales team. It lets users ask plain-English questions about sales pipeline, quota attainment, and rep performance, and get back governed, auditable answers grounded entirely in real data — every answer is required to show its SQL, raw data, a confidence score, and a trace ID. No hallucinated numbers, no black-box outputs. **What you'll be implementing (see attached spec for full detail):** - Raw CSV ingestion into cleaned, quarantine-validated Delta tables - Governed SQL semantic views for quota attainment, pace/risk scoring, and quarter-over-quarter comparisons - Row-level security by sales rep via Unity Catalog - A Genie-powered natural-language-to-SQL layer, with a trust layer, clarification gate, and scope guardrails - A structured, grounded answer-composition layer (LLM output verified against real data before it reaches the user) - Feedback capture and full audit logging on every request path - A golden-question test set plus a unit test suite - A Streamlit front end deployed via Databricks Apps - Observability, drift detection, and canary monitoring - CI-gated deployment via Databricks Asset Bundles **Tech stack required:** Databricks Unity Catalog, Delta Lake, Genie Spaces, Databricks SQL Warehouses, Foundation Model APIs, MLflow (Tracing + Evaluate), Databricks Apps, Databricks Asset Bundles, Lakehouse Monitoring, PySpark/Python. **What we're providing:** A complete implementation guide (attached PDF) with architecture, schemas, view definitions, function-level reference code, a build-order sequence, and an acceptance checklist. Your job is to build against this spec faithfully, flag anything that doesn't hold up against real data, and deliver a working, deployed system that passes the checklist at the end of the document. **Requirements:** - Proven hands-on experience with Databricks (Unity Catalog, Delta Lake, SQL Warehouses) — please share past project examples - Strong Python/PySpark skills - Experience with MLflow (tracing and/or evaluation) - Experience with LLM-backed or NL-to-SQL applications is a strong plus (Genie experience especially) - Comfortable owning a full build independently, from raw data to deployed app — not just individual tickets **Deliverable:** A fully implemented, tested, and deployed application matching every section of the attached specification, with the acceptance checklist in the final section completed and verified. Please review the attached document before bidding, and reference specific sections in your proposal to show you've read it.
Skills
Fuente original: freelancer