Big Data Architect (AI)

About the Company

We are one of the world's earliest and most established digital asset platforms, serving tens of millions of users across global markets. Consistently ranked among the top exchanges worldwide, we support a comprehensive suite of trading and wealth management products across thousands of digital assets. With thousands of employees spanning key financial hubs globally, we operate with a remote-first, high-autonomy culture that attracts talent who thrive in fast-moving, intellectually demanding environments. We are building the financial infrastructure of the next era of the internet.

The Role

Business users at every level need data — and right now, getting it still requires an engineer. This role exists to close that gap permanently. You'll lead the end-to-end production delivery of a Data Agent platform — Text-to-SQL, report generation, anomaly detection — built on top of a mature, layered data warehouse. In 12 months, you'll have shipped an Agent system that non-technical teams use daily, reduced analyst bottlenecks at scale, and built the evaluation framework that keeps it honest.

Responsibilities

• Own the full lifecycle of the Data Agent platform: semantic layer design, schema automation, RAG knowledge base, Tool Calling architecture, and Prompt engineering — from prototype to production
• Build and maintain the semantic layer that gives LLMs accurate warehouse context, ensuring SQL generation holds up against complex, ambiguous business queries
• Establish the Agent evaluation framework: SQL accuracy benchmarks, answer consistency metrics, regression detection — and own the results
• Partner directly with product, operations, and risk control teams to turn Agent capabilities into usable products, and to make data retrieval accessible without engineering involvement
• Set technical direction for the intersection of data warehousing and LLM application development — including framework choices, retrieval architecture, and toolchain evolution

Requirements

• Has built and shipped a production Data Agent or AI data assistant — not a prototype, not a demo: live, monitored, with real users and measurable SQL accuracy
• Has owned end-to-end warehouse architecture across the full modelling stack (ODS through ADS) and can reason about schema design in the context of LLM consumption, not just analytics
• Deep hands-on experience with at least three of: Hive, Spark, Flink, ClickHouse, Doris, Trino — with real SQL tuning work, not just configuration
• Solid Python engineering behind LLM application frameworks (LangChain, LlamaIndex, or equivalent) — has built agentic multi-step reasoning pipelines and debugged them in production

Nice to Have

• Background in a regulated industry — financial services, brokerage, or similar — where data governance and auditability are non-negotiable
• Has contributed to open-source data or AI projects with visible community impact