AI Solutions - Manager- Consulting

Location

Anywhere in Country

Manager, AI/ML Engineer – Memory Layer & Knowledge Graph

The Artificial Intelligence and Data team at EY uses cutting‑edge technology to deliver AI solutions to clients. As a manager, you will build the memory layer – graph databases, ontologies, vector stores, retrieval services and grounding pipelines – that power our agentic AI platform.

• Architect and build the end‑to‑end memory layer of EY’s cognitive harness.
• Design and implement knowledge graphs, ontologies, vector indices and hybrid retrieval services.
• Build and operate graph database infrastructure (Neo4j, Spanner Graph, Neptune, Tiger Graph, Stardog, …) including schema design, ingestion, query optimization and integration.
• Engineer vector and hybrid retrieval stacks, embeddings pipelines, reranking and lexical‑plus‑dense retrieval services.
• Build memory services for working, episodic, semantic, and procedural memory with TTL, retention, consolidation and provenance for regulated workloads (SOX, HIPAA, GDPR).
• Implement grounding pipelines connecting agent runtimes to the memory layer with low latency, citation tracking and hallucination guardrails.
• Lead a team of AI/ML and data engineers, set technical standards, conduct code reviews and mentor on production engineering rigor.
• Collaborate with data‑science leadership on memory representation, retrieval quality and on translating prototypes into hardened production services.
• Stay abreast of AI trends (new graph paradigms, embedding models, retrieval techniques, agents) and recommend tools and patterns that fit client ecosystems.
• Develop and execute target memory architectures that enable implementation, monitoring, and evolution of agentic AI at scale.

• Strong AI/ML engineering background owning a memory service from schema to API to deploy.
• Deep hands‑on knowledge of graph databases (Neo4j, Spanner Graph, Neptune, Tiger Graph, Stardog) and query languages (Cypher, GQL, SPARQL, Gremlin).
• Grasp of ontologies and knowledge‑representation vocabulary, schema rules, instances, axioms and provenance.
• Experience with vector databases and hybrid retrieval – embeddings, ANN indices, reranking, query rewriting and semantic caching.
• Solid software‑engineering fundamentals – Python (and ideally Java/Go/Type Script), API design, testing, CI/CD, containerization and observability.
• Experience designing and operating production AI systems on a major cloud (GCP, AWS, Azure, Databricks).
• Track record of leading engineering teams, setting technical direction, mentoring and delivering production systems on time.
• Excellent communication skills for explaining memory and graph concepts to clients and defending engineering decisions.

• Master’s preferred in Computer Science, Software Engineering, Data Engineering or a related field;
  Bachelor’s with strong applied experience also considered.
• 6+ years of applied AI/ML or data‑engineering experience, with at least 2 years leading engineering teams.
• Demonstrable production experience with graph databases (Neo4j, Spanner Graph, Neptune, Tiger Graph, Stardog, or similar), including schema design and query tuning.
• Hands‑on experience designing and implementing ontologies and knowledge graphs for real systems.
• Production experience building retrieval/RAG/memory systems – vector indices, hybrid retrieval, embedding pipelines, reranking and grounding.
• Strong proficiency in Python and the modern AI/ML stack (PyTorch/Tensor Flow, Hugging Face, Lang Chain/Lang Graph).
• Practical experience with at least one agent framework (Google ADK, Bedrock Agent Core, Lang Graph, Auto Gen, OpenAI Agents SDK) and one major cloud AI platform.
• Client‑facing or cross‑functional experience delivering production systems under enterprise constraints.

• Experience with semantic web standards (RDF, OWL, SHACL, SPARQL) and modern property‑graph approaches.
• Designing memory architectures that satisfy regulated‑industry constraints (SOX, HIPAA, GDPR) – retention, audit lineage, explainability.
• Experience with neuro‑symbolic patterns combining graphs and ontologies with LLM reasoning.
• Experience with…