Navy Jupiter

Platform Overview

Jupiter is the Department of the Navy enterprise data environment. It launched in April 2020 under the Department of the Navy Chief Information Officer (DON CIO) and Chief Data Officer (CDO), with a mandate to make DON data discoverable, accessible, understandable, and usable across the Naval enterprise.

The platform serves Navy and Marine Corps military personnel, civilians, and contractors. Structurally, Jupiter is the DON subtenant of Advana, the DoD-wide analytics platform managed by the Chief Digital and AI Office (CDAO). You get DON data sovereignty and Navy-specific governance, with Advana's enterprise tooling available underneath.

The platform's three core mission areas: Warfighting (readiness, maintenance, and operations data), Business (financial management, procurement, and contracts), and Readiness (personnel, training, and health).

Getting Access

The DON CDO released a policy memo establishing baseline access for all DON personnel: any Navy or Marine Corps military member, civilian, or contractor with a valid CAC or PIV token can access Jupiter and selected baseline datasets without additional approval steps.

You go to https://jupiter.data.mil, authenticate with your CAC or PIV, and you're in. This is not how most government data platforms work. Jupiter's baseline CAC access policy eliminates the provisioning friction that typically takes weeks.

For most DON data scientists starting a new project, the workflow is:

1. Authenticate at https://jupiter.data.mil with your CAC
2. Open the Collibra data catalog and search for relevant datasets
3. Identify whether the datasets you need are in a publicly accessible space or require elevated access
4. If elevated access is needed, contact the data steward listed in Collibra or reach DON_Data@navy.mil

Available Tools

iQuery is the DON-specific data query interface. It gives analysts direct access to DON data assets for ad hoc exploration and querying without requiring Databricks notebook infrastructure. SQL-based, fast for direct questions.

Collibra is the data governance and metadata platform. Every dataset entry includes metadata: the data source, update frequency, data owner and steward, applicable sensitivity designations, and which quality tier (bronze, silver, or gold) the data occupies. The standard practice for any new DON analytics project is to start in Collibra, not in code.

Databricks handles the heavy lifting for data engineering and data science. Python, PySpark, and SQL are all available in Databricks notebooks. If you are building ML models, running large-scale transformations, or engineering data pipelines, Databricks is your primary environment.

Qlik provides business intelligence and visual analytics for analysts who do not work in notebook environments.

Tableau rounds out the visualization toolkit. The CNO Executive Metrics Dashboard — built by NIWC Atlantic — is the highest-visibility production Tableau deployment on the platform.

Data Sources and Quality Tiers

Jupiter connects to over 63 DON source systems across warfighting, business, and readiness domains. Jupiter implements a three-tier data quality model:

Jupiter is also approved to process PII and PHI. That approval is uncommon enough to be worth calling out explicitly. Personnel analytics, health readiness modeling, and behavioral analysis against individual service member records are all within scope — with appropriate access controls.

Data Science Workflows

The standard path: catalog exploration in Collibra → data access and initial profiling in iQuery or a Databricks notebook → exploratory analysis against bronze or silver data → feature engineering and model development in Databricks against silver data → validation against gold-tier authoritative data → deployment back into a dashboard or reporting pipeline via Tableau or Qlik.

# Standard Jupiter data access pattern via Databricks
import pandas as pd
from pyspark.sql import SparkSession
from pyspark.sql.functions import col, datediff, current_date

spark = SparkSession.builder.appName("ReadinessAnalysis").getOrCreate()

# Access gold-tier maintenance data (requires appropriate access role)
maintenance_df = spark.table("gold.surface_force.maintenance_actions")

# Filter to open maintenance actions
open_actions = maintenance_df.filter(col("status") == "OPEN")

# Calculate days since scheduled completion
# Surface force data has known schema variation pre/post FY2019 — verify before joining
open_actions = open_actions.withColumn(
    "days_overdue",
    datediff(current_date(), col("scheduled_completion_date"))
)

overdue = open_actions.filter(col("days_overdue") > 30)

print(f"Open maintenance actions: {maintenance_df.count():,}")
print(f"Actions >30 days overdue: {overdue.count():,}")

# Feature engineering for predictive maintenance on surface ships
import numpy as np
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report

feature_cols = [
    "ship_age_years",
    "hull_type_code",
    "last_drydock_days_ago",
    "maintenance_backlog_count",
    "deployment_days_ytd",
    "avg_days_overdue_last_6mo"
]
target_col = "became_critical_within_90_days"

# Drop rows with NaN in features — maintenance data has known gaps
model_df = overdue_pd[feature_cols + [target_col]].dropna()

X = model_df[feature_cols]
y = model_df[target_col]

X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42, stratify=y
)

clf = RandomForestClassifier(n_estimators=100, max_depth=8, random_state=42)
clf.fit(X_train, y_train)

print(classification_report(y_test, clf.predict(X_test)))

# Feature importance — useful for explaining predictions to operators
importances = pd.Series(
    clf.feature_importances_, index=feature_cols
).sort_values(ascending=False)
print("\nFeature importances:")
print(importances)

Security and Compliance

Jupiter holds Authority to Operate on all three major DoD classification networks — NIPRNET, SIPRNET, and JWICS. Most government analytics platforms operate on a single network tier. Jupiter's three-network accreditation means a team working on personnel analytics at the unclassified level and a team working on operational intelligence analytics at TS/SCI are both working within the Jupiter ecosystem, with different data spaces and access controls appropriate to their classification tier.

Neptune Cloud Management Office

Jupiter runs inside infrastructure managed by the Neptune Cloud Management Office, established formally in 2023 under PEO Digital. Neptune is the DON's cloud management and delivery function — a "concierge service" for digital offerings across the enterprise. As a data scientist, you do not manage the underlying cloud infrastructure. Neptune handles that. Your responsibility stops at the Databricks notebook and the Collibra catalog.

Neptune Cloud Management Office organizational structure. Jupiter is one of four enterprise services managed under Neptune, with separate instances across the three DoD classification networks.

Key Use Cases

CNO Executive Metrics Dashboard

Built by Naval Information Warfare Center (NIWC) Atlantic and launched in January 2025, the dashboard feeds from Jupiter's gold-tier data and provides Admiral Franchetti with real-time readiness, manning, and maintenance metrics. The dashboard features over 12 automatically updated metrics accessible through clickable graphics, used weekly for briefings at the Pentagon, before Congress, and at the White House. Automated data pipelines from DON source systems through Jupiter's bronze-to-gold governance workflow, final visualization built in Tableau, updated automatically without manual data pulls. This is the template for what production analytics on Jupiter looks like at scale.

Task Force Hopper

Launched in summer 2021 and named for Admiral Grace Hopper, Task Force Hopper is the Naval Surface Force's AI/ML operationalization effort. The task force selected Advana-Jupiter as its primary common development environment for data storage, cleaning, model development, and deployment. The task force adopted a federated model — centralized catalog and standards enforced through Jupiter, decentralized AI development nodes at the fleet level where domain expertise exists. Use cases include predictive maintenance for surface ships, readiness analytics, and operational lethality improvement studies.

Navy Reserve

The U.S. Navy Reserve — approximately 59,000 personnel — adopted Jupiter formally in 2021. The Reserve established a Force Data Officer (FDO) role (filled by Capt. Kathleen Powell) to lead its data science program within the platform. The primary Reserve use case involves training budget optimization, with Jupiter providing the personnel records, training histories, and budget data in one place — including the PII approvals needed to work at the individual reservist level.

Relationship to Advana

Advana enterprise architecture showing Jupiter as the DON subtenant. Other service branches occupy parallel tenant spaces within the same enterprise structure.

Data scientists working in Jupiter can access both DON-specific data (the 63+ source systems) and DoD-wide data available through Advana's enterprise layer. If you need to join DON personnel data to a DoD-wide contract action database, the architecture supports it.

Where This Goes Wrong

Platform Close

The one thing to remember: Jupiter's value is not the tools — it is the governance structure that makes DON data trustworthy enough to inform decisions at the level of a CNO briefing. The bronze/silver/gold tiering and Collibra catalog are not process overhead; they are why gold-tier data can credibly appear in a Pentagon briefing. Use the governance infrastructure, not around it.

What to do Monday morning:

1. Authenticate at https://jupiter.data.mil with your CAC. Confirm your access tier.
2. Open Collibra and search for your domain. Spend two hours understanding what datasets exist.
3. Contact the data steward for your primary dataset — ask about known data quality issues, schema history, and update frequency.
4. Open a Databricks notebook and start with exploratory analysis against bronze or silver data.
5. For access help beyond the catalog, email DON_Data@navy.mil.