Data Scientist - National Federal Tax Services

Data Scientist role focused on building and optimizing LLM-powered prototypes and production-ready components using Generative AI and NLP. Responsibilities include designing prompts, implementing RAG workflows, packaging models as APIs, supporting cloud deployment on AWS/Azure/GCP, and developing evaluation and monitoring approaches for reliability and compliance. Requires strong Python proficiency, GenAI framework experience, and familiarity with software engineering best practices.

What you'd actually do

  1. Build and optimize GenAI/NLP models and prototypes using modern framework e.g., OpenAI, Hugging Face).
  2. Design, test and refine prompts to improve quality and reduce risk (e.g., hallucinations, bias, unsafe outputs).
  3. Implement foundational Retrieval-Augmented Generation pipelines to enable context-aware applications.
  4. Package models as APIs and support deployments on AWS (Amazon Web Services, Azure, or GCP (Google Cloud Platforms).
  5. Develop evaluation approaches and monitor model outputs for reliability, performance drift, and compliance needs.

Skills

Required

  • Python
  • GenAI frameworks (e.g., OpenAI, Hugging Face)
  • LangChain
  • OpenAI/GPT APIs
  • Hugging Face
  • Git
  • CI/CD
  • automated testing
  • R
  • SQL
  • cloud ML deployment (Azure, AWS, or GCP)

Nice to have

  • fine-tuning LLMs
  • building conversational AI agents
  • Responsible AI
  • privacy
  • security
  • AI ethics considerations
  • data engineering
  • MLOps

What the JD emphasized

  • strong Generative AI (GenAI) and Natural Language Processing (NLP) skill
  • design prompts and Retrieval-Augmented Generation (RAG) workflows
  • package models as APIs
  • cloud deployment
  • outputs are evaluated, monitored, and aligned to responsible AI expectations
  • hand-on Python development
  • solid statistical modeling fundamentals
  • 1+ year hands-on building LLM/NLP solutions
  • Strong Python proficiency
  • GenAI frameworks
  • Responsible AI
  • privacy, security, and AI ethics considerations

Other signals

  • LLM-powered prototypes
  • GenAI and NLP skill
  • design prompts and RAG workflows
  • package models as APIs
  • cloud deployment
  • responsible AI expectations
  • Python development
  • statistical modeling fundamentals