Systems Engineer, Query Execution

About the role

A plan is only as good as its execution. PLOMID turns one plan across rows, documents and events into running operators: vectorized where it pays, parallel where it is safe, bounded in memory always. This role builds that runtime: the code that turns intermediate results into answers without drowning the machine.

What you’ll do

• Implement vectorized execution operators over columnar and row data.
• Build parallel and incremental execution paths with explicit memory budgets per query.
• Work with the planner on operator interfaces: what an operator promises about ordering, batching and spilling.
• Measure relentlessly: add the benchmarks and profiles that keep optimization honest.
• Harden the edges operators usually get wrong: NULL semantics, type coercion, empty inputs, skew.

What we’re looking for

• Systems programming experience in Rust, C++ or a similar language, with attention to cache behavior and allocation.
• A working understanding of query execution: iterators versus batches, joins, aggregation, sorting.
• The habit of measuring before optimizing, and of writing down what was measured.
• Comfort working against interfaces owned by another layer (the planner) without blurring the boundary.

Nice to have

• Experience with columnar formats or analytical engines.
• Familiarity with SQL semantics beyond the common subset.
• Contributions to open-source databases or data processing tools.

What you’ll work on

The execution layer described on the architecture page: operators, parallelism, memory discipline. You will work most closely with whoever owns planning, because the contract between plan and operator is where execution quality is decided.

Why this role matters

Users judge a database by what comes back and how long it takes. The storage layer keeps the promise; execution keeps it on time. This role owns the second half.