Generative AI Applications Engineer (Agents & RAG)

Description

The company serves the US federal government across defense, national security, public safety, civilian, and military health. Its 13,000+ staff work on the goal of using technology to make the country safer and improve people's lives.

The team is a technology company inside global Accenture and was named a Glassdoor Top 100 Best Place to Work. It offers hands-on learning, certifications, and industry training in a collaborative setting.

The company builds production generative AI applications for confidential federal programs, where reliability, privacy, and safety are required. Releases happen in weeks, and success is measured by latency, reliability, safety, and cost.

Program details are not published. Specifics are shared with candidates who advance in the process.

This role turns mission needs into secure, reliable, scalable GenAI applications without training models. The work is hands-on and covers agentic workflows, retrieval-augmented generation (RAG), prompt and policy design, LLM evaluation, and platform integration. You will own the path from assessing a use case to production launch and ongoing operations, working with product, security, data, and SRE teams.

Responsibilities

• Build agentic workflows and RAG systems fitted to mission data and environments, aiming for low hallucination rates, tight p95 latency, and predictable cost.
• Use LangChain, LlamaIndex, or Semantic Kernel patterns to design task breakdown, tool use, guardrails, and recovery or fallback paths.
• Integrate AWS Bedrock, Azure OpenAI, Google Vertex AI, Amazon Kendra, and managed services such as Document AI, Gemini, and Gemma, without training models.
• Compare models on quality, safety, latency, and cost. Write and test prompts and policies, then deploy with observability and safe rollback or fallback.
• Build retrieval pipelines and vector search using Pinecone, Weaviate, OpenSearch, pgvector, FAISS, or Chroma. Handle data preparation, chunking, and metadata, and measure retrieval quality with IR-style evaluations such as NDCG.
• Instrument metrics, logs, and traces. Run A/B experiments, keep incident playbooks current, and add safety and compliance guardrails.
• Define SLIs and SLOs for quality, latency, safety, and cost. Take part in on-call rotations and postmortems, reduce mean time to recovery, and track token usage and spend.
• Build reusable SDKs, CI/CD templates, Terraform modules, and evaluation harnesses that serve several mission teams rather than single projects.
• Deliver into hybrid, restricted, or air-gapped environments using Zero Trust principles and audit-ready controls.

Requirements

• Have built and deployed a production GenAI application, such as a chatbot, copilot, assistant, or enterprise AI tool.
• Have worked hands-on with LLMs such as GPT, Claude, Llama, Gemini, or Mistral, through APIs or self-hosted.
• Have designed and built RAG solutions using embeddings, vector databases, and semantic search over enterprise or mission data.
• Have shipped AI applications with frameworks such as LangChain, LangGraph, LlamaIndex, Semantic Kernel, DSPy, or similar.
• Write strong Python for building and integrating AI and ML applications.
• Have owned AI solutions across the full production lifecycle, including deployment and operational support.
• Hold active US citizenship.
• Have experience integrating leading cloud AI services or on-prem inference stacks.
• Have background in LLM evaluation, prompt authoring and testing, A/B experimentation, and LLMOps.
• Understand responsible AI, including privacy, security, bias, transparency, human-in-the-loop design, and data governance.
• Have built tool-using agents for API integration and external data access.
• Use containers and orchestration (Docker, Kubernetes, VMware) and scripting or automation (Linux Bash, PowerShell).
• Have worked in regulated or secure environments, such as those with ATO, STIGs, or Zero Trust controls, while shipping quickly.
• Are familiar with NVIDIA AI Foundations, OpenAI ChatGPT, and AI-assisted development tools such as Cursor, Windsurf, and Claude.
• Have contributed to internal frameworks or open source, and have mentored engineers.
• Communicate clearly with engineers, product managers, and security or compliance stakeholders.

Conditions

Applications are accepted on an ongoing basis with no fixed deadline.

As required by local law, a base pay range is shown for the following locations: California, Colorado, Connecticut, Hawaii, Illinois, Maine, Maryland, Massachusetts, Minnesota, New Jersey, New York, Ohio, Vermont, Virginia, Washington, and the District of Columbia.

In those locations, the base pay range is $103,200 - $203,400 USD.

Actual pay depends on factors such as office location, role, skill set, and experience level.

The company provides equal opportunity and expects every employee to help keep an inclusive culture.

Benefits

• A wide range of benefits is offered, with details available separately.
• Hands-on experience, certifications, and industry training to support growth and learning.