Applied AI Engineer

J.D. Power•Ontario - Remote, ON

1d•$130,000 - $160,000•Remote

About The Position

As an Applied AI Engineer on the Launch Lab team, you will deliver full-stack AI-powered applications (internal tools, lightweight client-facing prototypes, and integration accelerators) on 2-to-4-week sprint cycles. You will take PoC blueprints from the Innovation Crew and own them through production: building interfaces and services, deploying them, confirming availability, and handing them off to maintaining teams. This role requires strong full-stack instincts, product intuition, and deep AI-native habits. Within the first 30 days, you will receive your first Blueprint Handoff and have a sprint plan drafted. By day 60, you will have a working application in use with real users. By day 90, you will have executed at least one clean handoff to a maintaining team. The Launch Lab builds on the Innovation Crew's agent infrastructure, consuming Power Agents and the Agent Skills Registry. You will surface Intelligence layer capabilities through production-quality interfaces and build eval harnesses and observability layers to ensure trustworthiness. Every application will have instrumented traces, a defined cost budget, and documented evaluation criteria before shipping. AI coding tools are used as a daily force-multiplier. The applications shipped will reduce manual hours, unblock product work, and demonstrate AI-powered delivery at startup velocity within the company. Your work directly enables JD Power's four capability pillars: AI Assistants and Copilots, Agentic Workflows, Engineering AI, and shared foundations.

Requirements

A portfolio of shipped, production AI applications: you have built and deployed real things real users depend on, can walk through what failed in production and how you fixed it, and treat "it worked in the demo" as the beginning of the work - not the end.
Full-stack engineering depth: frontend proficiency (React preferred) with the ability to produce polished production-quality interfaces at speed; backend depth across REST and/or GraphQL, authentication patterns, and cloud-native service architecture.
Hands-on experience with agentic frameworks or LLM APIs (LangChain, LlamaIndex, Anthropic, OpenAI, or equivalent), RAG patterns, streaming responses, and building interfaces that make AI behavior feel intuitive rather than opaque.

Nice To Haves

Eval harness design and LLM observability (Langfuse, LangSmith, Braintrust, or equivalent): you have built evaluation frameworks for AI features and can define what "quality" means for a non-deterministic system before writing a line of production code.
Multi-model routing and cost optimization (LiteLLM, Portkey, or equivalent); experience managing inference cost as a first-class engineering constraint.
Data platform experience (Snowflake, BigQuery, or equivalent) and security basics for AI applications (PII handling, secrets management, prompt injection defense, API audit logging).
Strong product intuition: you ask "what does the user actually need?" before choosing a model; you write design docs, run your own postmortems, and can explain a model failure to a non-technical stakeholder without panic.
JD Power internal platform, data infrastructure, or product ecosystem familiarity - institutional knowledge compresses your ramp and your day-one impact significantly.

Responsibilities

Lead application architecture and implementation across active Launch Lab delivery tracks: build AI-powered full-stack applications on 2-to-4-week sprint cycles, from wireframe through polished production-quality interface.
Support the frontend and UX quality of all Launch Lab outputs including adoption of a shared component library and design system.
Receive, interpret, and execute PoC Blueprint Handoffs from the Innovation Crew: translate validated architectures (Power Agents integrations, vector DB hooks, agent workflow patterns) into production applications with real users and measurable outcomes.
Build evaluation harnesses and observability instrumentation that make every application's AI behavior inspectable in production.
Build on the Innovation Arena's layered architecture: consume Power Agents and Agent Skills Registry modules; surface Intelligence layer capabilities (Data Product Layer, Industry Intelligence Agents, Domain Context Workflows) through polished user-facing applications.
Apply QE Agent Capabilities and Infrastructure and Tooling from the Quality and Delivery layer to maintain delivery standards.
Integrate Launch Lab outputs into JD Power's core data infrastructure: Snowflake, internal APIs, and the API Gateway - following published gateway standards and working closely with Internal Platforms to confirm data access and connection patterns.
Ensure AppSec, Data Governance, PII handling, and OEM data requirements are embedded in design, not bolted on after the fact.
Participate in Gate 2 technical scoping: scope new requests, estimate delivery effort, identify integration dependencies, flag Power Agents applicability, and define measurable success criteria including the evaluation criteria and cost budgets for each AI feature before building begins.
Own each Launch Lab output through to live: deploy in its target environment, confirm it is operational and available for users, then execute a clean handoff to the maintaining team within Product Engineering, Engineering, OEM Solutions, Infrastructure, or Internal Platforms - leaving behind documentation the receiving team can actually use.
Contribute application-layer technology evaluations to the Emerging Technology Radar; propose directional Architecture Decision Records for patterns and technology choices made during active delivery tracks.
Conduct lightweight user testing with business unit partners.