DSML Engineer – Generative AI

DSML Engineer – Generative AIAbout the Role

The successful candidate will help DSML teams move from fragmented discovery and manual project setup to a reusable, governed acceleration capability. The engagement is structured in two phases, with a formal continuation decision at the end of Phase 1.

What You'll Be Doing

Governed Knowledge and RAG

·Partner with data and business owners to onboard approved documents, data definitions, prior work, and expert knowledge into a governed knowledge base.

·Implement retrieval pipelines, metadata, taxonomy tagging, source citations, and quality checks that make context discoverable and trustworthy.

·Create evaluation datasets and retrieval-quality metrics, and diagnose gaps in relevance, grounding, completeness, and source stewardship.

·Ensure generated outputs clearly distinguish retrieved facts, inferences, assumptions, and items that require subject-matter-expert (SME) confirmation.

Agentic Product Engineering

·Context Agent: Build an agent that retrieves and organizes approved DSML context into structured, usable outputs for data scientists and stakeholders.

·DSML Use-Case Agent: Build an agent that translates business objectives, available data, and prior-project knowledge into structured data science and machine learning use cases, including requirements, assumptions, risks, and recommended next steps.

·DSML Project Scaffolding Agent: Develop an agent that converts a use case into a standardized project starting point, generating the repository structure, starter code, documentation templates, and workflow guidance needed to begin delivery efficiently.

·Design multi-step agent workflows that retrieve approved context, use tools safely, produce structured outputs, and retain human review points.

·Implement prompt, workflow, and tool-orchestration patterns that improve repeatability, traceability, and usability for data scientists, ML engineers, business analysts, product owners, and SMEs.

Evaluation, Delivery, and Governance

·Define and instrument technical and user-centered evaluation of agent outputs, covering correctness, completeness, traceability, revision effort, latency, and cost-to-serve.

·Work in two-week agile sprints: demonstrate working increments, document technical decisions, and convert pilot feedback into a prioritized backlog.

·Apply secure development practices and meet Responsible AI, security, privacy, data-governance, and enterprise review-board requirements.

·Partner with the future operating owner to define maintainable code, runbooks, monitoring, knowledge-source refresh practices, and an enhancement backlog.

What You Bring

·Bachelor's degree in Computer Science, Data Science, Machine Learning, or a related field, or equivalent practical experience.

·5 years of software, data science, machine learning, or AI engineering experience, including production-quality Python development and collaborative, version-controlled delivery.

·Hands-on experience building LLM-enabled applications, RAG systems, agentic workflows, or comparable AI assistants using APIs and structured tool integrations.

·Strong Python skills and experience with modern software engineering practices, including testing, code review, CI/CD, API design, documentation, and observability.

·Experience with data retrieval and indexing concepts, embeddings, vector or hybrid search, evaluation design, and quality measurement.

·Ability to turn ambiguous business problems into testable technical requirements and to communicate trade-offs to technical and non-technical partners.

·Demonstrated commitment to responsible AI, privacy, security, and data-governance practices.

Preferred Qualifications

·Experience with modern LLM orchestration, agent-workflow, or developer-assistance platforms.

·Experience designing LLM evaluation frameworks, guardrails, prompt-version management, telemetry, or model-risk controls.

·Experience supporting ML lifecycle workflows, such as feature or data documentation, model development, or ML project bootstrapping.

·Familiarity with data governance, responsible AI, access control, auditability, and human-in-the-loop review.