Senior AI Platform Engineer

 AI Platform Engineering – Core Orchestration & Runtime

Build the PAR Restaurant AI Platform - a dual-purpose system that at build time gives product engineering teams the APIs, SDKs, tool registry, and knowledgebase services to author and deploy agents, and at runtime hosts those agents in production across all PAR product lines with multi-tenant isolation, security guardrails, content safety, governance, and compliance controls

Design and build production-grade GenAI microservices (e.g. FastAPI) and stateful multi-agent workflows, including selection of the appropriate orchestration pattern (model-driven, graph-based etc.) based on auditability, latency, and complexity requirements of each agent class

Design and operate an automated content ingestion pipeline that includes content upload, chunking, embedding model selection, vector storage, and OpenSearch Serverless indexing - exposing knowledgebase retrieval as a fully managed, self-service capability for all product teams

AI Platform Engineering – Tooling & Developer Experience

Build and maintain the platform APIs and developer tooling, including MCP-compatible tool interfaces, real-time streaming APIs (SSE / WebSocket), and agent configuration SDKs, that allow engineers to create, deploy, run, and observe multi-agent workflows with configurable guardrails and content safety controls per agent class

Own internal AI SDK ergonomics for both audiences: agent-authoring engineers and application-backend engineers consuming the streaming invocation API

Build and maintain the Platform Onboarding Copilot (Slack-integrated); produce clear, actionable technical documentation and architectural decision records (ADRs); present platform strategy to non-technical stakeholders

Own the platform observability stack as a product, not an afterthought. Provides platform tools for engineers to fetch observability metrics, and distributed traces auto-provisioned on every agent deployment; an LLM eval suite (e.g. LangSmith, Ragas) that gates CI/CD on quality regressions; and per-tenant cost attribution with spend alerting giving every engineering team full visibility into their inference consumption

Define quality bars and eval suites per agent class; no agent reaches production without a documented, passing score threshold, thereby establish evaluation as a first-class engineering discipline across the team

Governance, Compliance, Multi-Tenancy

Design multi-tenant isolation from day one, including namespace isolation, tenant_id injection, and policy-as-code enforcement ensuring there is no cross-brand data or context pollution.

Own the policy-as-code library for PAR-specific policy sets covering PAR Restaurant product lines while ensuring platform compliance with SOC 2 Type II, PCI DSS, GDPR, and CCPA; author and maintain the GDPR deletion pipeline

Define and implement content safety guardrails: grounding checks and content filtering mandatory for all customer-facing agents

Act as the quality and cost-economics gatekeeper for all agents entering pre-production, such as running standardised eval suites, enforcing agent best practices, and validating inference cost-per-run against approved thresholds before any agent is cleared for production deployment

Collaborate with DevOps (owners of CI/CD automation) through a well-defined infrastructure contract and build version promotion pipeline (e.g. GitHub Actions)