It is if the petition isolates the company-specific elements she holds, such as the proprietary reconciliation rules and calibration logic she designed, and shows they are not readily available in the US labour market. General data-engineering skill is not enough, however advanced.
Where these petitions usually fail
The regulation's two limbs are special knowledge of the company's product, service, research, equipment, techniques or management, or an advanced level of knowledge of the company's processes and procedures. Petitions go wrong by arguing the engineer's talent rather than the company's specificity. A description that could be pasted into a job advertisement for a data engineer in any city proves the opposite of what is needed.
Instead, name the components: the matching algorithm that reconciles yield files from different monitor brands, the field-boundary logic, the rules for discarding suspect passes, the internal tooling nobody outside the company has seen. Explain who wrote them and how they are documented. Then explain the gap: the US subsidiary and its dealers can operate the product but cannot adapt its core logic without someone who knows why it was built that way.
Avoid claiming she is the only person who could ever do this; the test is whether the knowledge is specialized and hers, not whether she is irreplaceable.