Senior MLOps / AI Platform Engineer

Infopact•Arlington, VA

About The Position

Build, integrate, deploy, and operate the AWS-hosted chatbot and orchestration services, including its connections to Databricks, enterprise documents, vector stores, and participating websites.

Requirements

9+ years of experience in MLOps, cloud engineering, platform engineering, DevSecOps, or production AI application development.
Hands-on experience building and deploying AI/ML applications using the AWS technology stack, including Amazon Bedrock and/or SageMaker, API Gateway, ECS or EKS, ECR, S3, IAM, Secrets Manager, KMS, CloudWatch, and related services.
Strong experience containerizing Python-based applications and APIs using Docker.
Hands-on experience integrating applications with Databricks REST APIs.
Experience building API-based chatbot or agent-orchestration services using Python and frameworks such as FastAPI, Flask, LangChain, LangGraph, or comparable technologies.
Experience implementing routing across structured-data, RAG, and hybrid question-answering workflows.
Experience connecting AI applications to vector stores such as Amazon OpenSearch Serverless, Bedrock Knowledge Bases, PostgreSQL/pgvector, or Databricks Vector Search.
Experience developing document-ingestion pipelines for SharePoint, S3, file repositories, or other enterprise knowledge sources, including extraction, chunking, embeddings, metadata enrichment, synchronization, and deletion handling.
Experience implementing secure token handling, OAuth/OIDC flows, service-principal authentication, secrets management, least-privilege IAM, and server-side session management.
Experience creating CI/CD pipelines for containerized AI applications, including automated testing, vulnerability scanning, image promotion, deployment, rollback, and configuration management.
Experience load-testing and scaling chatbot or model-enabled applications based on concurrent users, request volume, model latency, token utilization, and downstream API constraints.
Experience implementing production observability, including application logging, distributed correlation IDs, metrics, tracing, API latency monitoring, model usage monitoring, alerting, and audit-log integration.
Ability to troubleshoot issues spanning application code, Docker containers, AWS networking, IAM, Databricks APIs, SQL execution, model endpoints, and vector retrieval.
Ability to implement guardrails that limit data returned to the model, restrict the application to approved Databricks views, prevent credentials from reaching the browser, and preserve end-to-end auditability.

Nice To Haves

Experience deploying applications within the War Data Platform, formerly Advana, or integrating with WDP-hosted Databricks services.
Experience supporting AWS GovCloud, DoD IL4/IL5, CUI, or other highly regulated environments.
Experience with Databricks Genie, Unity Catalog, Databricks Vector Search/AI Search, and user-level OAuth integrations.
Experience integrating reusable chatbot widgets or APIs into multiple websites, such as PBIS, Jupiter Homepage, Community Portal, or similar enterprise applications.
Experience with Terraform, CloudFormation, AWS CDK, Helm, Kubernetes, and Git-based CI/CD platforms.

Responsibilities

Build, integrate, deploy, and operate AWS-hosted chatbot and orchestration services.
Connect services to Databricks, enterprise documents, vector stores, and participating websites.
Containerize Python-based applications and APIs using Docker.
Deploy containerized applications into scalable ECS, EKS, or Kubernetes environments.
Integrate applications with Databricks REST APIs.
Build API-based chatbot or agent-orchestration services using Python and frameworks like FastAPI, Flask, LangChain, LangGraph, or comparable technologies.
Implement routing across structured-data, RAG, and hybrid question-answering workflows.
Connect AI applications to vector stores.
Develop document-ingestion pipelines for enterprise knowledge sources.
Implement secure token handling, OAuth/OIDC flows, service-principal authentication, secrets management, least-privilege IAM, and server-side session management.
Create CI/CD pipelines for containerized AI applications.
Load-test and scale chatbot or model-enabled applications.
Implement production observability, including logging, tracing, metrics, and alerting.
Troubleshoot issues spanning application code, Docker containers, AWS networking, IAM, Databricks APIs, SQL execution, model endpoints, and vector retrieval.
Implement guardrails to limit data returned to the model, restrict the application to approved Databricks views, prevent credentials from reaching the browser, and preserve end-to-end auditability.