GraphDB Architect

Job description

Key Responsibilities • Architecture ownership — define the end-to-end solution architecture across Ingest, Entity Resolution Engine and Serve workspaces on Microsoft Fabric; produce solution blueprints, architecture diagrams and integration patterns. • Entity ontology & canonical model — design the entity ontology, canonical data model, attribute and provenance model, and the identifier spine • Entity resolution strategy — define the matching approach: deterministic rules on shared identifiers, blocking strategy for candidate generation, probabilistic scoring features, confidence banding and survivorship rules. • Graph architecture — design the graph schema (nodes, edges, properties), define relational-to-graph projection logic, model entity/ownership/affiliation relationships, and design incremental re-projection on CDC. • Platform decisions — evaluate and select the graph and vector platform approach (Fabric-native Graph vs alternatives), with a supporting capacity, performance and cost model. • Performance & capacity design — design Spark pool configuration and workspace/capacity strategy for compute-intensive resolution workloads; optimise Delta file sizes, partitioning and pipeline efficiency. • Standards & governance — define data access policies (RBAC, Fabric security roles), data quality rules, metadata standards and naming/versioning conventions. • Design assurance — review deliverables including code, models, pipelines and graph schemas; mentor engineers and ensure alignment to architectural standards. • Client engagement — present and defend architecture decisions to WK stakeholders and Microsoft; support technical discovery and design workshops. Required Skills & Experience Skill Area Specific Requirements Graph Technology Labeled property graph (LPG) modelling, graph query languages (GQL / Cypher / Gremlin), multi-hop traversal design, graph schema design, projection patterns MDM Deterministic and probabilistic matching, blocking strategies, survivorship and golden-record design, corporate hierarchy modelling, identifier spines Microsoft Fabric Lakehouse, Warehouse, OneLake, Data Factory, Spark/notebooks, Mirroring & CDC, capacity and workspace design, Fabric Graph Architecture Medallion / multi-zone lakehouse (Bronze→Silver→Gold), data modelling (dimensional, Data Vault), integration patterns, API design AI / Retrieval GraphRAG concepts, vector search and embeddings, natural-language-to-query approaches Engineering Depth Python/PySpark, SQL, Delta Lake, performance tuning, distributed compute optimisation Leadership Design authority, technical mentoring, client-facing architecture presentation, trade-off analysis and decision documentation Must-Have Qualifications • 12+ years in data engineering/architecture with at least 3 years designing graph or MDM solutions • Hands-on architecture experience with graph databases and graph data modelling • Demonstrable entity resolution / record linkage design experience at scale • Deep Microsoft Fabric or equivalent modern lakehouse platform expertise • Experience owning architecture decisions in a client-facing enterprise engagement • Strong hands-on ability — this is a working architect role, not advisory only Nice-to-Have • Experience with Splink or comparable probabilistic linkage frameworks • Exposure to GraphRAG or retrieval-augmented generation over knowledge graphs • Background in corporate/legal entity, KYC, credit or compliance data domains • Microsoft certifications (Fabric Analytics Engineer, Azure Solutions Architect) Key Deliverables Owned • Solution architecture and design documentation (HLD/LLD) • Entity ontology and canonical data model • Entity resolution strategy: match features, blocking and threshold design • Graph schema and node/edge projection logic • Graph & vector platform decision paper with capacity/cost model • Architecture and data-flow diagrams; reusable component standards Dual Role / Complementary Skills Strong complementary overlap with the Entity Resolution strategy — this architect is expected to define the matching approach that the data engineering team implements. Can also act as interim technical lead for the Graph and Vector engineers during ramp-up, and is the natural escalation point for performance and capacity issues.