AIP Innovation Engineer - iDEA by Lear

About the role

The AIP Innovation Engineer is a hands-on builder and visionary to demonstrate "what is possible" with AIP/AI/Agentic AI across existing and new Foundry solutions. You'll design, implement, and operationalize LLM/agent workflows, integrate internal and external data sources, and partner with Ontology Leads to shape data for maximum automation.

Responsibilities

• Design and implement AIP agents and LLM-backed workflows (prompt flows, tools, skills, policies) that are grounded in Foundry Ontology objects and feature sets.
• Leverage and stay current on Palantir’s platform innovations (e.g., AI FDE, AI Pilot), bringing forward the right capabilities at the right time.
• Data & Integration Engineering:
  • Identify opportunities via hands-on data work and analysis to drive necessary harmonization/transforms, and semantic grounding to support AI/LLM-based solutions.
  • Identify internal and external integrations (partner data, supplier feeds, SaaS apps) with appropriate security, throttling, and resilience patterns to bring the right data together, securely to leverage AI.
• Ontology Driven AI:
  • Partner with Ontology Leaders to propose and refine ontology objects, relationships, and reusable semantics that unlock automation and cross-use case reuse.
  • Influence data shaping required for RAG/grounding, action execution, reasoning chains, and multi-agent handoffs.
• Reliability, Data Health & Guardrails:
  • Collaborate with Data Quality Lead for intelligent monitoring: data quality checks, freshness, schema drift, lineage, latency SLOs, evals for LLM output quality, "red team" prompts, and safety guardrails.
  • Establish evaluation harnesses (offline/online) for agent workflows and prompts; track regression metrics and cost/performance KPIs.
• Productionization & Performance:
  • Build CI/CD pipelines, IaC where applicable, and observability (logs, traces, metrics) for AIP agents to ensure AIP agents and Foundry applications operate reliably at scale and can be quickly diagnosed, tuned, and improved.
• Solution Delivery & Stakeholder Collaboration:
  • Work closely with product owners, plant operations, quality, supply chain, and finance to scope high-value use cases; rapidly deliver MVPs and iterate to scale.
  • Provide clear technical documentation, runbooks, and handoffs to operations teams.

Qualifications

• 4+ years building production data/AI solutions (startup or enterprise); demonstrated hands-on ownership from ingestion to deployment.
• Strong experience with LLM/agentic systems: prompt design, tool/function calling, retrieval/grounding, safety policies, and evaluation.
• Proficiency with at least two of: Python, TypeScript/JavaScript, PySpark; comfort with APIs, microservices, and event-driven patterns.
• Experience with Palantir Foundry and/or AIP (Ontology, pipelines, transformations, apps, agents). If not Palantir, deep experience with adjacent stacks (e.g., LangChain/LangGraph/CrewAI/AutoGen/Semantic Kernel; vector DBs; cloud AI services) and the ability to ramp to Palantir quickly.
• Practical Data Quality & Observability experience (contracts, schema checks, lineage, alerts, evals) and a bias toward operational excellence.
• Comfortable working without a mature EDW—able to roll up sleeves to wrangle messy data, define interim schemas, and harden pipelines.

Preferred Qualifications

• Prior work integrating AI into manufacturing/industrial contexts (e.g., mapping to ISA-95 hierarchies, OEE, quality/NCR, routings, genealogy).
• LLMOps/MLOps experience (MLflow, model registries, eval pipelines, CI/CD for prompts/agents).
• Cloud experience (Azure/AWS) for scaling inference, storage, and data movement.
• Familiarity with secure by design patterns: identity, access, secrets, PII handling, audit logging.

Nice to Have Experience (Translatable if not Palantir)

• Built agentic AI systems using LangChain, LangGraph, CrewAI, AutoGen, Semantic Kernel.
• Implemented RAG with hybrid retrieval, adaptive chunking, and domain-specific guardrails.
• Designed distributed fine-tuning (e.g., QLoRA, instruction tuning) and stood up LLMOps/MLOps pipelines (MLflow, K8s, SageMaker, Ray).
• Delivered document intelligence (multimodal parsing, extraction, validation) and operational AI (recommendation, anomaly detection, forecasting).