Robotics Data Engineer

Askari Defense•Atlanta, GA

About The Position

This role is designed for a robotics data engineer who can build and own the data foundation behind our autonomy, perception, GNC, flight sciences, and test efforts. Data is the lifeblood of our engineering effort. We are a deeply data-driven team: we evaluate algorithms, characterize hardware, troubleshoot in the field, and train the models behind our autonomy by leaning hard on the data our robotic systems generate. We generate an enormous amount of it. Our field-testing tempo is high and our vehicles evolve quickly, producing a constant stream of new data that never stops growing. Our engineers already extract real value from this data every day. What we need now is to improve the systems that take a fast-growing volume of mission data across many sources and formats and turn it into a single, synchronized, queryable asset the whole company can build on. You will build and own that foundation: a unified data backbone that serves as the company’s source of truth across both high-rate time-series and imagery data. You will be responsible for the entire data lifecycle, from logging onboard the vehicle, through transfer, storage, and organization, to the analysis, tooling, and model-training pipelines that let every engineering team self-serve the cross-section of data they need. This is a high-ownership role for someone who can bring order and cohesion to a demanding data landscape, then mature it into robust, automated infrastructure the whole company depends on. The problem has wrinkles particular to our domain. A single mission can involve multiple vehicles, interceptor and target tracks, and increasingly swarms, producing parallel, heterogeneous data streams that must be synchronized and made coherent before they mean anything. Our hardware and software also evolve rapidly, so data must be versioned by configuration to keep comparisons across tests trustworthy. The fundamentals are shared with any high-performance UAS program, but doing this well against multi-vehicle, multi-entity flight data is where the real challenge lives.

Requirements

Degree in Computer Science, Robotics, or a related technical field, or equivalent practical experience.
A few years of professional experience in a data engineering, robotics, or backend software role, building systems that others depend on.
Strong proficiency in Python and its data ecosystem, including Pandas and NumPy, with the software discipline to build maintainable libraries rather than one-off scripts.
Proven experience structuring and manipulating large, complex datasets, especially time-series data drawn from multiple unsynchronized sources.
Ability to travel as needed to support field and flight-test operations.

Nice To Haves

Working knowledge of C++, sufficient to develop and maintain onboard logging software.
Hands-on experience with data formats common in robotics and data engineering, including uLog, ROS 2 bags, MCAP, Parquet, Protobuf, HDF5, or related formats.
Strong skills in data visualization for analysis and reporting, including high-quality programmatic plots using Matplotlib, Plotly, Seaborn, or related tools.
Experience building interactive, exploratory frontends that help engineers navigate and make sense of complex mission data, including Plotly Dash, Bokeh, Streamlit, Rerun, Foxglove, PlotJuggler, or related tools.
Experience with relational databases, such as PostgreSQL, and SQL for building queryable data stores.
Experience with cloud storage and its access patterns, including GCS buckets, object metadata, IAM, or related systems.
Experience building and maintaining CI/CD pipelines for data processing and automated testing.
Familiarity with communication protocols used to move data between systems, including ZMQ, WebSocket, or related protocols.
Familiarity with low-level Linux storage and I/O internals, including ext4, memory-mapped I/O, page cache, io_uring, NVMe, zero-copy, fsync/durability, and the ROS 2 stack used for onboard logging.
Familiarity with MLOps principles and tools for dataset versioning, experiment tracking, and training-pipeline automation.
Comfort scripting against off-the-shelf VLMs, including Gemini, Llama, or related models, for tasks such as data labeling and curation.

Responsibilities

Own the data lifecycle end to end, from what gets logged onboard the vehicle to the tooling engineers use to make sense of it.
Design and build the unified data backbone that ingests, parses, validates, and synchronizes data across multiple vehicles, sources, and modalities, aligning time-series telemetry and imagery into a single coherent timeline.
Develop a warehousing strategy that converts raw logs into efficient, queryable formats serving as the company’s source of truth for all mission data.
Manage and scale storage for the large datasets generated by flight testing.
Track data provenance and version it against the software and hardware configuration it came from, so analyses can filter by flight type, vehicle configuration, sensor payload, and software version while avoiding the mixing of incompatible data.
Partner with autonomy on, or take full ownership of, hardening the onboard C++ logging framework for efficiency, reliability, and completeness.
Develop and maintain core libraries for programmatic, offline-first data analysis: the primary way engineers load mission data, run analysis, and generate consistent, high-quality visualizations for reports, debugging, and development.
Build automated validation and regression pipelines integrated with CI/CD, flagging performance deviations automatically once mission data lands.
Mature existing tooling into a fully featured suite so engineers can compose analysis tasks without writing ad hoc code.
Build retrieval and query systems on top of the backbone, letting teams pull the exact cross-section of data they need by flight context, software/hardware version, or content.
Work directly with GNC, perception, flight sciences, and state estimation engineers to understand their data needs and build the routines that characterize system performance and troubleshoot issues.
Build the data infrastructure that feeds model training and evaluation, including versioning, labeling, and curating datasets so they are ready for ML pipelines.
Take on complex, data-heavy analysis tasks directly, including the backlog the team has already identified.