Data Architect

Role Title: Data Architect

Location: London, UK (Hybrid )

Job type: Contract

Job Description:

Data product architecture experience is mandatory - talent must have architecture experience in Data products

• Streaming & EDA: Kafka (Confluent) and AWS MSK/Kinesis/Kinesis Firehose; outbox, ordering, replay, exactly/at-least-once semantics. EventBridge for event routing and filtering.
• Schema management: Avro/Protobuf + Schema Registry (compatibility, subject strategy, evolution).
• AWS data stack: S3/Glue/Athena, Redshift, Step Functions, Lambda; Iceberg-ready lakehouse patterns. Kinesis→S3→Glue streaming pipelines; Glue Streaming; DLQ patterns.
• Payments & ISO 20022: PAIN/PACS/CAMT, lifecycle modeling, reconciliation/advices; API/File/SWIFT channel knowledge.
• Governance: Data-mesh mindset; ownership, quality SLAs, access, retention, lineage.
• Observability & FinOps: Build dashboards, alerts, and cost KPIs; troubleshoot lag/throughput at scale.
• Delivery: Production code, performance profiling, code reviews, automated tests, secure by design.

Data Architecture Fundamentals (Must-Have)

• Logical data modeling: Entity-relationship diagrams, normalization (1NF through Boyce-Codd/BCNF), denormalization trade-offs; identify functional dependencies and key anomalies.
• Physical data modeling:Table design, partitioning strategies, indexes; SCD types; dimensional vs. transactional schemas; storage patterns for OLTP vs. analytics.
• Normalization & design: Normalize to 3NF/BCNF for OLTP; understand when to denormalize for queries; trade-offs between 3NF, Data Vault, and star schemas.
• - CQRS (Command Query Responsibility Segregation):Separate read/write models; event sourcing and state reconstruction; eventual consistency patterns; when CQRS is justified vs. overkill.
• Event-Driven Architecture (EDA): Event-first design; aggregate boundaries and invariants; publish/subscribe patterns; saga orchestration; idempotency and at-least-once delivery.
• Bounded contexts & domain modeling: Core/supporting/generic subdomains; context maps (anti-corruption layers, shared kernel, conformist, published language); ubiquitous language.
• Entities, value objects & repositories: Domain entity identity; immutability for value objects; repository abstraction over persistence; temporal/versioned records.
• Domain events & contracts: Schema versioning (Avro/Protobuf); backward/forward compatibility; event replay; mapping domain events to Kafka topics and Aurora tables.

Nice-to-Have

QuickSight/Tableau; Redshift tuning; ksqlDB/Flink; Aurora Postgres internals.

- Edge/API constraints (Apigee/API-GW), mTLS/webhook patterns.