Senior Data Engineer

The Central Data Team (CDT) is only a few months old, but data has been Keyrock's lifeblood since day one. We're now building the Keyrock Data Platform to give Keyrockers, and the AI agents working alongside them, the data and context they need to act fast and autonomously within agreed boundaries and aligned with our shared goals. You'd be one of the early hires in CDT and we expect you to contribute to most of the bigger decisions and build work.

Responsibilities

Build streaming and batch pipelines that ingest, normalise, and distribute market, trading, and portfolio data, resilient to feed and exchange failures; Build the self-serve tooling (SDKs, patterns, templates, AI agents) so other teams publish, consume, and build on data products without waiting on us; Own data contracts and schema evolution. Keep schema changes from turning into multi-team coordination events; Design the lakehouse and time-series layer around consumer query patterns; Build and evolve the Data Governance and Data Quality Framework: stale-feed detection, schema validation, range checks, idempotent writes, lineage, ownership, self-healing; Build the derived analytics the business runs on: cross-exchange spreads, VWAP at depth, order book microstructure for the desks; portfolio views, exposure, performance for wealth and asset management; Make observability, cost, and performance first-class from day one; Treat infrastructure as code (Docker, Terraform, CI/CD) alongside our Central Infrastructure Team; Work in the open: write things down, partner closely with Architecture, Infrastructure, Platform, and the rest of the teams.

Requirements

8+ years of building production data systems that other people rely on; Strong proficiency in Python and SQL: not just being able to write a query, but being able to reason about what the engine is doing with it; Code that's easy for someone else to read, test, and delete later; Strong understanding of data modelling for both streaming and analytical workloads; Efficiency, quality, idempotency, and observability are taken seriously by default; You've designed and operated streaming systems on Kafka, Redpanda, MSK, or Kinesis, and you have opinions about partitioning, consumer groups, offsets, and schema registries; You've used a time-series store in production (ClickHouse ideally; TimescaleDB, QuestDB, or similar are fine too) and can talk about table design as a function of query patterns; You've worked with a lakehouse architecture and reason about table layout, partitioning, and compaction as design choices that shape query performance and storage cost; You build for self-healing and idempotency. Reprocessing is safe, retries don't double-write, and the system recovers without a human in the loop; Docker, Terraform, and CI/CD are how you work, not a separate "DevOps" thing; You think about cost and performance early; You instrument as you build: logs, metrics, and traces are part of the system from day one; You design for data quality and governance up front covering contracts, validation, lineage, and ownership; You reason from first principles when a problem is new, stay pragmatic when it isn't, and update your view when you learn more; You treat the trading desks, wealth and asset management, product, risk, finance, compliance, and research as customers of what you build, and communicate with them that way; You optimise for outcomes over output. A smaller, simpler thing that ships and works beats a bigger thing that doesn't; You take ownership end-to-end: design, ship, operate, improve; You say what you think including when it's an unpopular take. You change your mind when the argument is better; You make the people around you better. Reviews are real, juniors grow from working with you, and peers want to work with you again; You're curious about how markets work. Data engineering on its own won't keep you interested here; You're honest about what you know and what you don't, and quick to close the gap; You understand financial market data: order books, trades, reference data, portfolios, exposures. Crypto, TradFi, or both are a strong plus

Nice to have

Lakehouse experience with Apache Iceberg or Delta Lake; Familiarity with DataHub or similar metadata/lineage platforms; Rust. Some of our performance-critical services are written in it. Interest is welcome; fluency isn't required.

Benefits

From-scratch mandate as one of the first CDT hires; Strong partners in Architecture, Infrastructure, Platform, and desks; Autonomy on how you work; Flexible hours; Remote-first; Business-hours on-call shared across the team; Competitive salary package with various benefits; A team that likes each other; Regular online get-togethers; Yearly onsite