Generative AI Platform Engineer

Accenture Federal Services • Washington, DC

5h

About The Position

At Accenture Federal Services, nothing matters more than helping the US federal government make the nation stronger and safer and life better for people. Our 13,000+ people are united in a shared purpose to pursue the limitless potential of technology and ingenuity for clients across defense, national security, public safety, civilian, and military health organizations.
Join Accenture Federal Services, a technology company within global Accenture. Recognized as a Glassdoor Top 100 Best Place to Work, we offer a collaborative and caring community where you feel like you belong and are empowered to grow, learn and thrive through hands-on experience, certifications, industry training and more.
Join us to drive positive, lasting change that moves missions and the government forward!
Build AI that matters. At Accenture Federal Services (AFS), we turn GenAI from demos into operational capabilities that millions rely on. We ship production GenAI apps for confidential federal programs across defense, national security, public safety, civilian, and military health where reliability, privacy, and safety aren’t optional.
AFS is a technology company within global Accenture and a Glassdoor Top 100 Best Place to Work. You’ll join a collaborative, caring community that invests in your growth through hands‑on experience, certifications, and industry training. We ship in weeks, not quarters, and measure success by latency, reliability, safety, and cost.
Confidentiality matters: We don’t disclose program details on public postings. If you advance, we’ll share specifics as appropriate during the process.
Role Overview
Operationalize LLM/RAG: Stand up and harden AI services (cloud or on‑prem) that ground responses in mission data with low hallucination rates.
Guardrails & governance: Implement prompt/version management, policy‑based filtering, role‑based access, audit trails, and safety checks aligned to federal guidelines.
Observability & SRE for AI: Define SLIs/SLOs (quality, latency, safety, cost), instrument metrics, run on‑call, lead postmortems, and reduce MTTR.
Performance & cost engineering: Optimize p95 latency with caching/batching/autoscaling; meter usage and apply FinOps to keep token/spend within targets.
Reusable platform components: Ship SDKs, CI/CD templates, and Terraform/IaC modules that accelerate multiple mission teams not one‑off projects.
Applied theory (where it matters): Use IR metrics (e.g., NDCG) to improve retrieval quality; apply queueing/control principles to hit latency SLOs; run offline/online evals for helpfulness, safety, and factuality.
What you won’t do: Train or invent new foundation models; deliver academic research. Your focus is making best‑in‑class models work flawlessly in production.

Requirements

• Own production systems end‑to‑end: Integration → deployment → observability → incident response.
• Are strong in Python
• Have experience with retrieval/vector search (e.g., pgvector/Milvus/OpenSearch) and grounding AI in enterprise data.
• Think in SLOs and dashboards; you improve reliability, latency, safety, and cost with data.
• Communicate clearly with engineers, PMs, and security stakeholders especially in compliance‑heavy environments.
• Must be a US Citizen

Nice To Haves

• Experience integrating leading cloud AI services or on‑prem inference stacks
• Background in evaluation/safety testing or A/B experimentation for AI experiences.
• FinOps for AI: metering, budgets, and optimization strategies for usage and spend.
• Prior work in regulated/secure environments (e.g., FedRAMP‑like controls, ATO processes) with fast shipping.
• Contributions to internal frameworks or open‑source tooling; mentorship of engineers.
• Advanced degree welcome but not required high‑impact delivery matters most.

Responsibilities

• Operationalize LLM/RAG: Stand up and harden AI services (cloud or on‑prem) that ground responses in mission data with low hallucination rates.
• Guardrails & governance: Implement prompt/version management, policy‑based filtering, role‑based access, audit trails, and safety checks aligned to federal guidelines.
• Observability & SRE for AI: Define SLIs/SLOs (quality, latency, safety, cost), instrument metrics, run on‑call, lead postmortems, and reduce MTTR.
• Performance & cost engineering: Optimize p95 latency with caching/batching/autoscaling; meter usage and apply FinOps to keep token/spend within targets.
• Reusable platform components: Ship SDKs, CI/CD templates, and Terraform/IaC modules that accelerate multiple mission teams not one‑off projects.
• Applied theory (where it matters): Use IR metrics (e.g., NDCG) to improve retrieval quality; apply queueing/control principles to hit latency SLOs; run offline/online evals for helpfulness, safety, and factuality.

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level
• Education Level: No Education Listed
• Number of Employees: 1,001-5,000 employees