gravity9 is recruiting Java engineers and architects for a portfolio of engagements across UK central government. These are large-scale, data-intensive modernisation programmes: replacing legacy systems that carry statutory weight, decomposing monoliths that serve millions of users, and building secure data platforms where access control and auditability are first-class architectural concerns rather than late additions.
We are hiring across two overlapping profiles — Senior Java Developer and Application / Migration Architect — and expect most people to move between programmes as engagements start and complete. Several of these are hands-on architect roles: you will own design decisions and also write production code

The kind of work
The following are representative of current and upcoming engagements. Individual roles will draw on some of this, not all of it — we are more interested in engineers who can work confidently at the intersection of Java and data than in a checklist match against any one programme.
Secure data catalogue and access-control platform
A greenfield metadata catalogue acting as a federated index over data held across independent domains, so that authorised parties can locate and retrieve data the catalogue itself never holds. The work includes designing an attribute-based access control model over attributes such as clearance, nationality, organisational unit and compartment; enforcing those policies inside the database query path rather than filtering results afterwards; application-layer encryption with externally managed keys; hybrid lexical and semantic search with access filtering applied before candidate retrieval rather than after it; and a tamper-evident audit trail that can reconstruct why a given record was returned, redacted or withheld.
Real-time decision-support data architecture
Replacing point-to-point integration across a multi-vendor estate with a microservices architecture and a common data layer, so that operational data flows in near real time instead of over days. The work includes target-state architecture and decision records, change-data-capture ingest pipelines from upstream data services, ontology and mapping-layer design across suppliers with competing data models, separation of system, reference and insight data classes including time-series, and deployment topologies spanning cloud-hosted services down to constrained edge hardware.
Legacy-to-document-model migration at national scale
Migrating hundreds of millions of records and tens of terabytes of XML out of a legacy relational platform into a sharded document store, in support of a COBOL-to-Java microservices rewrite. The work includes a configuration-driven XML-to-JSON transformation framework backed by a versioned schema registry — twenty years of history, multiple record types and continual schema and semantic drift; CDC and streaming pipelines, or a checkpointed parallel bulk-extract alternative; a field-level mapping validation and reconciliation framework robust enough to stand up to audit; a replay and failure-diagnostics harness with dead-letter routing and idempotent re-runs; and repeated dress rehearsals toward a single cutover window with no dual running.
Monolith decomposition and data-ownership migration
Breaking a distributed monolith of well over a hundred Java services sharing a handful of database clusters into domain-owned services — moving from borrowed data to owned data without reintroducing hidden coupling. The work includes bounded-context analysis to establish which domain genuinely owns which collections, in-place data remodelling ahead of any physical move, a data-access facade with feature-flagged per-request routing between old and new stores, dual-write / backfill / reconcile / read-switch cutover lifecycles with a defined rollback at every step, cluster-to-cluster streaming and live migration tooling, and automated parallel-run and diff-on-read testing.

What you will be doing
Emphasis varies by role and programme, but MongoDB is the common thread. Every one of these engagements involves modelling for it, migrating onto it, or building on top of it, and genuine depth here is what most differentiates candidates.
Architecture and design
•      Produce high- and low-level designs, security and data architectures, and architecture decision records — and take them through client design authorities and formal governance gates
•      Make and defend architectural trade-offs in front of both engineering and non-technical stakeholders, including client technical authorities and third-party suppliers
•      Analyse real access patterns and validate target MongoDB document models against genuine throughput, query and sharding requirements before they are baselined
•      Design integration boundaries in multi-vendor programmes where you own one component and must collaborate across the seams of others
Hands-on build
•      Build production Java services and RESTful APIs (Spring Boot or similar), designed for integration by multiple consuming systems and documented to OpenAPI
•      Design and implement MongoDB document models, aggregation pipelines, indexing and sharding strategies, and integrate them cleanly into Java service layers rather than through a leaky abstraction
•      Build data pipelines: CDC and streaming ingest, transformation frameworks, checkpointed batch extract workers, and MongoDB sink configuration and tuning at terabyte scale
•      Implement access control, encryption and audit within the application and data layer, including enforcement inside the MongoDB query path, rather than bolted on around it
•      Write the tests: unit, integration, golden-record fixtures, parallel-run reconciliation, and load and stress harnesses that prove performance under realistic volumes
•      Work across both MongoDB Atlas and Enterprise Advanced, including Kubernetes-hosted deployments, and with the migration and sync tooling around them
Data modernisation a