Mayur Reddy

3D Foundation Models · SDK & API Development · Platform Architecture · Product Integration

• Drive architecture and development of reusable SDKs, APIs, and versioned data contracts for generative 3D workflows, creating stable developer interfaces between foundation-model pipelines and downstream design products.
• Design and ship shared translation and integration components that convert generated 3D outputs into validated, product-compatible representations, enabling reusable model-to-product workflows across teams and applications.
• Define API and platform architecture across ML research, platform, and product teams, including schema evolution, compatibility, validation, versioned artifacts, identity/provenance, and reliable handoff patterns for shared developer workflows.

3D Foundation Models · ML Infrastructure · Distributed Data Systems

• Built and operated petabyte-scale 3D data pipelines that transformed complex product data into training- and evaluation-ready datasets for foundation-model research, large-scale ML experimentation, and downstream processing on AWS.
• Developed a reusable Ray-based distributed processing framework for heterogeneous 3D workloads, isolating parser failures, bounding concurrency, checkpointing completed work, and producing durable outputs for recoverable large-scale processing.
• Built dataset cataloging, lineage, and provenance infrastructure tracing source assets through transformations and ML derivatives, improving reproducibility, dataset discovery, and experimental traceability.
• Built browser-based 3D streaming and inspection infrastructure for large models, enabling labeling, evaluation, and ML inspection workflows without requiring clients to download complete source datasets.