Real-Time Break Reconciliation Platform

I am building an intraday front- and back-office reconciliation intelligence platform that surfaces trade breaks within minutes instead of waiting for the usual end-of-day batch reports. The goal is twofold: first, to highlight exceptions in near real-time so operations teams can act while a position is still fresh; second, to recognise when multiple apparent breaks share one underlying root-cause and roll those into a single, actionable incident. Integration with SimCorp Dimension is central: the system will consume positions, cash ledgers, transactions, and instrument reference data through its published REST/OpenAPI services and supporting data-access interfaces, on both scheduled and event-driven cadences. A normalization layer maps SimCorp entities onto a canonical internal model, with authentication, throttling, and replay/backfill handling in scope so downstream detection logic stays vendor-neutral.Domain knowledge in Fund management, stock exchanges, mutual funds trading is a must. To do this, the solution must ingest and correlate data from the following sources: • Market data feeds • Trading systems • Risk management systems Low-latency processing, event-driven architecture, and a lightweight front end for alerts and drill-down analysis will be essential. If you have hands-on experience with streaming engines such as Kafka, Flink, Spark Streaming, or similar, coupled with solid knowledge of post-trade workflows, I’d like to see how you would structure the data pipelines, reconciliation logic, and incident-detection heuristics. Deliverables should include: • A working prototype able to process sample trade files and flag breaks within minutes • Logic to group related breaks under a single incident identifier • Simple dashboard or API endpoint that shows current breaks and their inferred root cause Code quality, clarity of documentation, and demonstrable performance in a near real-time test will be the key acceptance criteria.