Neural Data Infrastructure Engineer

Blackrock Neurotech•Salt Lake City, UT

55d•Onsite

About The Position

The Neural Data Infrastructure Engineer owns the systems that transform intracortical recordings into reliable, training-ready data for AI/ML models while preserving the scientific meaning of each recording. This hands-on individual contributor is the technical authority for the data lifecycle, from upload through training consumption, and works closely with model researchers, infrastructure and IT teams, and data owners. The role has end-to-end technical ownership of a growing intracortical BCI data corpus that supports Blackrock's neural foundation model work. The position is based on-site at Blackrock Neurotech's headquarters in Salt Lake City, Utah, with occasional travel potentially required. It is a high-ownership role, but the company states that it is not an “always-on” role and values sustained, high-quality work and life outside work.

Requirements

Bachelor’s degree in Computer Science, Electrical Engineering, Biomedical Engineering, Neuroscience, or a related technical field, or equivalent practical experience with 3+ years of relevant industry experience; alternatively, a Master's degree with 1+ years of relevant industry experience, or a PhD in a related field.
Demonstrated experience independently designing and operating large-scale scientific data pipelines end to end, with ownership of software quality, reliability, and delivery.
Exceptional programming ability in Python, with strong software design, debugging, profiling, automated testing, and version control practices.
Advanced understanding of extracellular neural recordings, spike detection and sorting, sampling theory, filtering, aliasing, referencing, and artifact handling.
Hands-on experience with intracortical electrophysiology data and its on-disk representation, including binary layouts, event timestamps, channel metadata, and continuous signals.
Ability to reason about how preprocessing choices alter neural information and affect model training, evaluation, and reproducibility.
Experience with scientific formats and storage systems such as NEV/NSx, NWB, HDF5, and Zarr, including efficient partial reads and metadata management.
Strong judgment about dataset lineage, validation, access permissions, and handling sensitive research data.
Ability to work across neuroscience, machine learning, and infrastructure teams and bring clarity to ambiguous requirements.

Nice To Haves

Experience with distributed or parallel processing, object storage, workflow orchestration, and restartable jobs at scale.
Familiarity with deep learning models and architectures, including how data processing decisions affect model training, inference, and evaluation.
Experience processing local field potentials, synchronizing neural and behavioral streams, or supporting longitudinal neural datasets.
Experience with compiled languages, high-throughput I/O, cloud cost management, or performance-sensitive scientific computing.
Fluency with deep learning data interfaces, including tensor shapes, batching, masking, sampling, and PyTorch.

Responsibilities

Own the design, implementation, and operation of the neural data pipeline from upload and validation through preprocessing, dataset releases, and training data loaders.
Build reliable ingestion for heterogeneous recording formats while preserving raw data, acquisition metadata, channel and electrode mappings, timestamps, units, and behavioral alignment.
Implement and validate neural signal processing for spike events and waveforms, including detection or sorting where needed, quality assessment, binning, and normalization; extend the pipeline to field potentials as the program evolves.
Establish automated checks for corrupt files, missing channels, clock drift, artifacts, recording discontinuities, and inconsistent metadata, with clear criteria for quarantine and recovery.
Design storage layouts, indexing, chunking, compression, and caching for efficient access to large neural datasets across local and cloud resources.
Create reproducible, versioned datasets with traceable transformations, provenance, and subject and session partitions that prevent leakage between training and evaluation.
Partner with model researchers to define input representations, sequence construction, masks, sampling policies, and interfaces that preserve neural meaning and meet architecture requirements.
Build and profile parallel preprocessing and streaming data loaders, partnering with the training performance engineer to keep GPUs supplied as training scales.
Work with infrastructure, IT, and data owners to plan capacity, access controls, retention, backup, and recovery appropriate to the recordings and their approved uses.
Establish testing, monitoring, documentation, and operational practices that make pipeline failures diagnosable and dataset releases dependable.
Communicate data quality, coverage, processing tradeoffs, and resource needs clearly, and translate evolving research requirements into maintainable engineering plans.