Senior Associate - Foundry​/AIP Forward Deployed Engineer

Senior Associate Engineer

Andersen Consulting is building a Palantir practice capable of competing with the largest system integrators on the planet. The practice will deploy Foundry and AIP across financial services, healthcare, supply chain, and manufacturing at enterprise scale, across global clients, starting now.

The Senior Associate Engineers who join at this stage will deliver within the practice. In addition to client facing work, you'll use Foundry and AIP internally across Andersen. Foundry and AIP sit at the center of how this firm delivers data and AI transformation. Senior engineers who join now shape the reference architecture every future engagement inherits.

You'll work with in Andersen Consulting's global AI practice, accountable for delivering measurable value to clients—both internally and externally—through data.

What You'll Do

• Build and maintain Foundry data pipelines in Pipeline Builder and Transforms, ingesting from structured, semi-structured, and API sources, handling schema drift, and pushing clean objects into the Ontology.
• Develop Workshop applications that surface operational data to business users: configuring widgets, wiring function-backed objects, and building action types that write back to source systems.
• Instrument AIP Logic workflows: prompting, chaining, and evaluating LLM-backed logic steps grounded in Ontology objects, with measurable pass/fail criteria before anything touches production.
• Contribute to AIP Agent builds: scoping agent tasks, wiring tool calls to Foundry functions, and testing failure modes against real client data in AIP Agent Studio.
• Write production-quality Python and SQL in Code Workbook and Code Repositories: shared transforms with unit tests that teammates can extend, not one-off notebooks.
• Sit with users. Run deskside sessions. Identify the actual decision points operators face before you write a single widget. Your designs should target what someone needs to approve, escalate, or prioritize, not just display.
• Participate in client discovery sessions to surface data availability and business process gaps before they become mid-engagement blockers.
• Build internal accelerators on bench time: reusable pipeline templates, Ontology design patterns, Workshop component libraries that become practice IP.

What You'll Build

• Ontology schemas: object type definitions, link types, and property type governance for client datasets across insurance, manufacturing, or healthcare domains.
• Pipeline architectures: end-to-end raw to bronze to silver to gold data models using Foundry Transforms, with documented lineage and observable data quality checks.
• AIP Logic modules: prompt-chained reasoning workflows grounded on Ontology objects, tested against structured evaluation sets before client demo.
• Workshop operations centers: dashboards wired to live Ontology objects, with writeback action types that update upstream systems and close decision loops.
• Notepad reports: automated narrative outputs that surface model results and pipeline summaries to non-technical stakeholders.
• Reusable accelerators: generalized pipeline templates and Workshop component patterns the practice deploys on future engagements without rebuilding from zero.

The Requirements

• 3+ years of production Foundry experience. You have shipped Ontologies, pipelines, and Workshop applications that real users depend on.
• Proficiency in Foundry's data layer:
  Pipeline Builder, Transforms (PySpark, Polars, and SQL), Object Storage, and Ontology design including link types and property type taxonomy.
• Working knowledge of AIP Logic; at least one prompt chain grounded on Ontology objects in production. You understand how action types surface to end users.
• Production-quality Python: write code that passes review, not just code that runs. You know when to use Polars vs. PySpark and why.
• Ability to break down ambiguous business problems into Foundry-solvable components. You translate a messy process into object types, properties, and action types before writing code.
• Comfort with organizational data as it actually exists: inconsistent identifiers, overlapping source systems, conflicting definitions.
• Data literacy as a client…