About this role
When supply chains fracture due to shortages or end-of-life components, manufacturers rely on Fusion Worldwide to keep operations moving. This business depends entirely on accurate intelligence regarding active companies, part substitutions, and demand, which is managed through an internal ERP and production data platform. The senior data engineer joining the core data platform team will solve the challenge of maintaining and scaling this architecture, where AI agents write the majority of code and connect directly to the platform for agentic development.
In this role, the engineer will build and manage ingestion pipelines, design APIs for internal UIs, handle system of record write-backs, and construct agent workflows that triage inquiries and flag anomalies. Because agents frequently fail in subtle ways—such as dropped rows in window functions or silent script failures—this person must review AI-generated code, enforce rigorous guardrails, and maintain strict data lineage. Success requires hands-on backend and data engineering expertise, a pragmatic approach to failure, and the ability to build explainable systems where parameters are versioned and outputs include clear audit trails.
Operating within a light Agile process alongside a product manager, the right candidate takes features from concept to production independently. The team values direct communication, rigorous documentation, and practical problem-solving over self-promotion. Anyone who has deployed LLM systems to production, encountered real-world failures, and adapted their workflow accordingly will thrive in this environment.
What you will do
- Build and maintain object models covering companies, parts, offers, and demand signals by adding properties and relationships as required by the data.
- Develop and oversee data ingestion pipelines from the internal ERP, Azure SQL, HubSpot, and vendor APIs alongside quality gates and freshness checks.
- Design and version APIs for the Web RMS user interfaces while ensuring response times meet application demands.
- Construct agent workflows that read the object model to triage inbound requests, identify pricing signals, and flag anomalies.
- Manage write-back paths into the system of record while handling transactional integrity, idempotency, and reconciliation.
- Tune performance across query plans, index designs, materialization strategies, and caching layers.
- Participate in incident response and on-call rotations when pipelines or write-back mechanisms fail.
What you need
- Bring over eight years of software building experience with a strong focus on backend and data engineering.
- Demonstrate expert SQL skills and deep experience with relational databases such as SQL Server, including reading execution plans and designing indexes.
- Possess production data platform experience using environments such as Databricks, Snowflake, Spark, or dbt.
- Show proven experience writing back into systems of record with attention to transactional integrity and conflict handling.
- Have designed APIs for external applications, managing versioning and UI requirements without breaking integrations.
- Deliver features end-to-end from initial concept through proof of concept, building, iteration, and release.
- Build and ship production LLM systems with established guardrails, evaluations, cost monitoring, and latency controls.
- Utilize AI agents daily on production systems while demonstrating awareness of where they assist and where they fail.
Nice to have
- Three or more years of hands-on experience with a production data platform and object modeling
- Python and PySpark proficiency
- Dimensional modeling and schema design judgment
- Entity resolution, master data, taxonomies, or knowledge graphs
- React, streaming ingestion, or ERP integration experience
Skills and tech stack
Questions candidates ask
What is the tech stack used for data and AI development?
The data stack relies on Microsoft SQL Server, Azure SQL, Python, PySpark, SQL, and TypeScript hosted on Microsoft Azure. For AI development, the team connects Claude directly to the platform for agentic workflows, alongside custom React applications and Atlassian tools for work management.
How do AI agents fit into the daily workflow?
Agents write most of the code, running transforms, querying data, and executing audits. Engineers on the team build and maintain the agents, establish their instructions and tool connections, review their output, and handle the checks that catch mistakes.
How are features taken from idea to release?
Engineers work alongside a product manager on direction and priorities, doing most of the scoping, building proofs of concept, iterating, and deciding when a feature ships. Work is tracked in Jira using a light Agile process with story point estimates.
What are the expectations around system explainability?
Everything built must be explainable so traders can trace numbers back to their source. Parameters and thresholds are stored as versioned data, and outputs generated by LLMs include plain-English reasons with direct links to source fields.
The application is on the official Fusion Worldwide careers page.
Apply on company site