My client is an early-stage, VC-backed financial services research firm using AI to transform how professional investors do research.
We're helping them identify an early core engineer to build their research data platform from the ground up — everything from getting raw information into the system through to the AI-powered tools our end users rely on daily. You'll have real end-to-end ownership of the technical build, not just a slice of it.
What you'll work on:
- Pulling in data from many different sources and formats, both clean and messy
- Working with real-world source documents that don't come pre-formatted — scans, inconsistent layouts, tables, content in more than one language
- Setting up and maintaining the data warehouse that everything else runs on
- Keeping pipelines running reliably — scheduling, monitoring, catching bad data before it causes problems
- Building the AI layer itself: retrieval-based systems, structured outputs, and ways to measure whether the AI is actually performing well
- Building the APIs and services that let other tools and teammates query the data
- Running on modern cloud infrastructure (GCP or AWS)
What we're looking for:
- Somewhere in the 5–10 year range of hands-on engineering experience, ideally having taken systems from idea to production more than once
- Fluent in Python and SQL — this isn't a "picked it up along the way" requirement, we need someone genuinely strong here
- Real experience wrangling unstructured or semi-structured documents into usable data
- A couple of years of hands-on work actually shipping LLM-based features, not just experimenting with them
- Comfortable with at least one modern orchestration tool (Airflow, Prefect, Dagster, or similar) and a cloud-native warehouse
- Having done this kind of "build it from scratch" role before, ideally at another early-stage company, is a big plus
Bonus Points:
- Have built data infrastructure specifically for AI/language model use cases
- Comfortable with dbt or similar transformation tooling
- Understand how to design for multi-tenant data access and permissions
- Some exposure to graph databases (Neo4j or similar)
Lin Wee
License No. 22C1076 | EA Reg: R1878551