• Design, build, and operate the ingestion systems that process large volumes of multimodal data into usable, well-structured datasets • Own the ingestion path end to end, from how data lands to how it is validated, processed, tracked, and made available downstream • Build modality-specific processing steps for real-world source data, such as medical imaging processing, audio and video metadata extraction, quality validation, and notes processing • Build parsers, validators, and normalization logic that can systematically handle messy, non-standard, and high-variance source formats • Turn repeated one-off data handling work into reusable processing patterns, internal tooling, and platform capabilities • Build for high volume and high throughput, optimizing systems for reliability, cost, and speed • Work across distributed and parallel compute systems to process workloads that do not fit well on a single machine • Choose the right execution model for the workload, including batch processing, distributed execution, and modern compute patterns for unstructured data and inference-heavy processing • Diagnose and resolve bottlenecks across ingestion and processing systems, and keep performance from degrading as volume and modality complexity grow • Build validation and quality checks that catch bad, incomplete, or malformed data before it propagates downstream • Handle sensitive and regulated data, including PHI, with the security and care the domain demands, including de-identification where required • Track provenance, metadata, and usage constraints through the ingestion path so downstream use remains compliant and auditable • Raise the quality bar for observability, debuggability, and operational reliability across the ingestion layer • Partner with product and Data Lab to support new modalities, new partner requirements, and non-standard source data • Work directly with partner engineering teams when needed to translate source-system realities into robust ingestion and processing design • Surface recurring patterns that are worth standardizing into reusable transforms, validators, and internal tooling • Help shape how Protege handles new data types as the platform expands into more complex data environments