Sr. Application Solution Architect

Owning the end-to-end solution architecture of the Daitics AI CDP: a sovereign, on-prem, telco-native Customer Data Platform built on a streaming architecture, deployed on infrastructure we run ourselves (Kubernetes, Helm), not managed cloud services. The platform processes hundreds of thousands of events per second across thirty or more source systems and delivers unified customer profiles for B2C persons and B2B organizations, accounts, sites, lines, devices and contacts. You will own the architecture across the five planes (Authoring, Control, Data, Activation, Observability) and the five layers (atomic events, tile primitives, trait values, signal filters, signal events), hold the line on architectural principles across engine boundaries, and translate them into designs engineering teams can build against.

Requirements

• 12+ years in software and data engineering, with at least 5 years owning solution or platform architecture for large-scale distributed systems

• Demonstrated ownership of end-to-end architecture for a streaming data platform on self-managed infrastructure (on-prem or private cloud) topology, state placement, failure domains, cutover, and capacity, not documentation alone

• Deep, hands-on architectural command of Apache Flink as a stateful streaming runtime DataStream and SQL, keyed state on RocksDB, broadcast state for configuration distribution, Async I/O for external lookups, checkpoint and savepoint discipline, and failure isolation across independent job clusters (core requirement)

• Proven ability to design two-tier latency architectures: a millisecond-level hot path against a minute-level durable cold path, with an explicit end-to-end latency budget and a documented inclusion and exclusion list

• Experience designing Kafka topologies for multi-tenant platforms topic taxonomy, partition-key selection and partition sizing, compacted control streams, retention classes, and enforced naming conventions

• Experience architecting schema governance across producers and consumers Avro or equivalent through a schema registry, compatibility modes, and coordinated promotion of breaking changes

• Hands-on architecture experience with lakehouse table formats (Apache Paimon, Iceberg, Delta Lake, or Hudi) on S3-compatible object storage — primary-key and LSM table design, partitioning, compaction, snapshot expiry, and separation of streaming from batch read paths

• Experience designing governed single-writer patterns where many producers write to one substrate through one consolidating service, with per-stream configuration applied without restarts and exactly-once-effective semantics

• Experience with distributed in-memory caches (Apache Ignite, Hazelcast, Redis, or similar) partitioned versus replicated tables, affinity colocation with the compute layer, thin-client access patterns, operator-local caching, and cross-consumer invalidation

• Ability to define and enforce a state-placement decision framework: broadcast state versus keyed state versus external cache versus first-seen load, with explicit rules rather than case-by-case judgment

• Experience designing aggregation architectures that serve arbitrary rolling windows with sub-millisecond reads while keeping state bounded and recomputable (pre-aggregated tiles, multi-resolution granularity ladders, or equivalent), including approximate aggregation through sketches (HyperLogLog, t-digest, count-min) and honesty about their limits

• Experience architecting identity resolution at scale deterministic and probabilistic matching, identity graph modeling, effective-dated B2B relationships, merge and split workflows, anonymous-to-known stitching, and downstream propagation of identity change

• Experience modeling entity lifecycle state machines in streaming systems, including identifier reassignment, quarantine, archival, and purge, with auditable transitions

• Architecture experience with consent enforcement and data governance configurable enforcement modes, data usage labeling and enforcement, policy distribution to the runtime, and enforcement designs that avoid a synchronous service call per event

• Experience architecting tokenization, hashing, masking, and encryption for sensitive identifiers — reversible versus irreversible choices, key versioning and rotation (HashiCorp Vault Transit or equivalent), protection applied at the ingest boundary, and audited just-in-time detokenization at delivery

• Experience designing end-to-end lineage and edit-time impact analysis artifacts and dependencies modeled as a graph, forward and reverse traversal, and prevention of orphaned dependencies on delete, rename, or semantic change

• Experience designing versioned artifact activation and cutover across engine boundaries grace windows, transition emission policy, late-event windows, and rollback

• Ability to reason about state reconciliation when a computation definition changes forward-only, full or partial recalculation, parallel versions, evacuate and rebuild, snapshot seeding and how data retention constrains which options are valid

• Experience designing multi-tenant isolation on Kubernetes namespace and resource topology per tenant and workspace, noisy-neighbour control, per-tenant storage, secrets, and cross-workspace sharing contracts

• Experience architecting authentication, authorization, and network security enterprise IdP federation (OIDC, Keycloak or equivalent), attribute-based access control, Kubernetes NetworkPolicies for east-west isolation, service-mesh mTLS, and secrets management

• Experience defining failure domains, SLOs, RPO and RTO targets for streaming platforms, and designing replay, backfill, and reconciliation paths that do not re-trigger external actions

• Ability to size a platform of this shape end to end: from events per second and subject counts to partitions, parallelism, keyed state size, cache footprint, and storage — and to validate the model against measured behaviour

• Judgement on when to wrap an open-source component behind a stable platform service interface with pluggable implementations, and when to depend on it directly

• Experience with a canonical or industry semantic model and the discipline of preventing semantic drift; telco experience with TM Forum SID and TMF620 is a strong advantage

• Experience integrating ML into a streaming platform models upstream of derived attributes, online and batch scoring paths, training-serving consistency, and keeping inference off the hot path unless benchmarked

• Practical experience with Spring Boot on Java for control-plane microservices, PostgreSQL as a definition store, and workflow orchestration (Temporal, Argo Workflows, or equivalent) with BPMN approval flows (Flowable, Camunda, or equivalent)

• Experience with observability stacks (OpenTelemetry, Prometheus, Grafana, distributed tracing) and with defining what operators actually need to see

• Familiarity with graph and vector substrates for lineage, relationship traversal, and semantic similarity over platform metadata

• Ability to author and maintain a layered architecture corpus a master blueprint with component annexes, explicit design rules, and cross-document consistency and to defend it in architecture review

• Familiarity using AI tools for design, development, and debugging (Claude, Cursor, Codex)