> Markdown version of [/jobs/ext/4623400-senior-data-scientist-gen-ai](https://www.wearedevelopers.com/jobs/ext/4623400-senior-data-scientist-gen-ai). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Senior Data Scientist + Gen AI

- **Company:** AIT Global, Inc.
- **Location:** Mountain View, CA, United States
- **Experience:** Expert
- **Contract:** Contract
- **Skills:** Artificial Intelligence, Airflow, Amazon Web Services, Confluence, JIRA, Microsoft Azure, Cloud Computing, Python (Programming Language), Machine Learning, Role-Based Access Control, OpenAI, Management of Software Versions, Workflow Management Systems, Retrieval-Augmented Generation, Large Language Models, Prompt Engineering, Hallucination Detection, Llamaindex, Generative AI, Agentic-AI, Low Latency, Drift Detection
- **Published:** November 7, 2026
- **Apply:** https://www.dice.com/job-detail/b17a2057-81e3-4729-9aaf-4a03e43d57a4

## About the Role

* Large Language Models, AWS, Prompt Engineering, Python, Machine Learning, Gen AI, Confluence, Jira

## Description

Prompt engineering (system prompts, tool prompts, function calling, versioning with evaluations)
Evaluation & observability, including:

 * Ground truth setup
 * Confusion metrics
 * Hallucination rate measurement
 * LLM-as-judge with human review
 * Cost & latency monitoring
 * Retrieval strategies & prompt patterns
 * Context management
 * Hallucination mitigation
 * Faithfulness & grounding validation
 * Agentic workflows
 * Multi-step reasoning
 * Tool orchestration
 * Agent success rate measurement
 * LangChain / LlamaIndex (or equivalent) proficiency
 * Cloud LLM providers
 * Azure OpenAI
 * AWS Bedrock
 * Vertex AI
 * Workflow orchestration
 * Airflow
 * Dagster
 * Security & privacy
 * PII handling
 * RBAC
 * Strong Python expertise with real production experience
 * Success Criteria
 * Production-ready prompt + RAG system
 * Measurable quality lift (accuracy, faithfulness, confusion metrics)
 * Reduced hallucinations
 * Repeatable evaluation harness
 * Observable system (quality, latency, cost, drift monitoring)