#4558 NASA JSC Workforce Functional Taxonomy; +Generative-AI and Agentic-AI Applications Used By
#4558 NASA JSC Workforce Functional Taxonomy; +Generative-AI and Agentic-AI Applications Used By
At a federal, mission-driven installation like NASA’s Johnson Space Center (JSC), human capital is organized through a highly rigorous, multi-tiered framework. Because JSC acts as the hub for human spaceflight operations, its workforce functional taxonomy must reconcile standard Federal Civil Service structures with dense engineering and mission-critical lifecycles.
JSC cross-references its workforce using two primary taxonomies: the NASA Supplemental Classification System (NSCS) for civil service personnel, and Mission Directorates/Directorate Functional Nodes for operational execution.
1. The NASA Supplemental Classification System (NSCS)
To map specific technical competency requirements far more precisely than standard government-wide Office of Personnel Management (OPM) codes allow, NASA utilizes the NSCS.
At JSC, the workforce is primarily bucketed into these functional domains:
The 700 Group – Professional Engineering & Scientific (AST): This is the core technical engine of JSC.
Positions in this group are designated with the acronym AST (Aerospace Technology). This taxonomy explicitly ties an individual’s classification to precise technical specialties (e.g., AST-Flight Systems Safety, AST-Life Support and Habitation Systems, AST-Mission Specialist Astronaut). This group covers professional research, vehicle design, and real-time flight operations. The 600 Group – Professional Administrative: Positions requiring deep analytical judgment and administrative methodology to run the enterprise, including Procurement, Legal Counsel, Human Capital Management (OHCM), and Safety & Mission Assurance (S&MA) process compliance auditors.
The 500 Group – Clerical/Assistant Support: Personnel managing established administrative procedures, document control, and information systems.
The 300 Group – Technical and Nonprofessional Support:
Technicians and specialists utilizing practical, hands-on application of equipment, hardware fabrication, and laboratory infrastructure management.
2. Directorate Functional Nodes (Operational Architecture)
When looking at how the taxonomy of "Employees, Management, Development, Operations, Training, Planning" actually manifests on the ground at JSC, the workforce is structured across functional Directorates.
Every individual—whether Civil Servant (CS) or Engineering Services Contractor (ESC)—is mapped to a node that dictates their position in the mission lifecycle:
[ JSC Center Leadership / Planning ]
│
┌───────────┴────────────────┐
▼ ▼
[ Development & Engineering ] [ Operations & Training ]
├── Engineering Directorate (EA) ├── Flight Operations Directorate (FOD)
└── Safety & Mission Assurance (NA) └── Exploration Architecture (XA)
Planning & Governance
Center Leadership & Program Offices: Taxonomically isolated to manage strategic planning, budget formulation, and inter-agency agreements (e.g., Orion, Gateway, Commercial Crew, and ISS Program Offices). They bridge the gap between NASA HQ policy and JSC execution.
Development (The "Build & Certify" Track)
Engineering Directorate (EA): The individual contributors and managers tasked with designing, prototyping, and testing hardware/software systems. This is where structural, propulsion, avionics, and software process engineers reside.
Safety and Mission Assurance (S&MA - NA): The independent oversight body. The taxonomy here is heavily structured around safety engineers, quality assurance specialists, and software assurance professionals who verify that design metrics map perfectly to NASA technical standards (like NPR 7150.2D).
Operations, Training, & Execution (The "Fly" Track)
Flight Operations Directorate (FOD - CA): This directorate handles the real-time execution and crew readiness tracks. It bifurcates into distinct functional paths:
Flight Directors & Flight Controllers: The operational teams in the Mission Control Center (MCC) responsible for continuous vehicle tracking, systems monitoring, and real-time command execution.
Astronaut Office: Mission Specialists and Pilots categorized as the operational users of the developed platforms.
Training & Simulation Infrastructure: The personnel tasked with building human-in-the-loop training blocks, high-fidelity mockups, and software simulations (such as SVMF—Space Vehicle Mockup Facility) to prepare crews for anomalous flight conditions.
3. The 17 Technical Disciplines (Cross-Cutting Technology Taxonomy)
Finally, for data-mining, skill-tracking, and project matching within JSC's TechPort and internal knowledge graphs, the workforce's specific expertise is categorized under the broader NASA Technology Taxonomy (TX).
Because JSC owns the human spaceflight portfolio, its engineering and operations workforce predominantly classifies its technical output under:
TX02: Flight Computing and Avionics
TX06: Human Health, Life Support, and Habitation Systems (ECLSS, EVA, HSI)
TX10: Autonomous Systems (Reasoning, Acting, and Human-Robot Interaction)
TX11: Software, Modeling, Simulation, and Information Processing
Are you looking to align this specific JSC taxonomy with an internal systems architecture model, or are you looking for a deeper breakdown of the AST 700 Group codes for software safety and quality tracking?
List the generative-AI and agentic-AI applications that groups 300, 500, 600 and 700 likely use
When mapping Generative-AI (GenAI) and Agentic-AI applications to the NASA Supplemental Classification System (NSCS) at Johnson Space Center, the tools segment clearly by risk tolerance, operational domain, and technical complexity.
Because NASA mandates rigorous data-layer governance to prevent data exfiltration, the tools used range from highly secure, federally approved commercial instances to sandboxed, local neural/symbolic hybrid models.
The 700 Group: Professional Engineering & Scientific (AST)
High-fidelity, deterministic, and highly governed environments.
The AST group deals with Class A flight software, systems engineering (SysML v2), and safety-critical domains. They rely heavily on hybrid neural-symbolic systems where generative insights are verified by physics-based boundaries or mathematical logic engines.
Generative-AI Applications
Model-Based Systems Engineering (MBSE) Assistants: Tools like the AI for Systems Engineering (AISE) assistant, which uses LLMs to parse free-form engineering queries (asymmetric queries) and cross-reference them with complex SysML v2 repository models to promote validated design reuse.
Automated Code Analysis & Boilerplate Generation: Internal, secure code-generation environments (similar to sandboxed GitHub Copilot Enterprise) optimized for C/C++ and compliant with strict aerospace standards like the Core Flight System (cFS) framework.
Technical Standard Synthesis: RAG (Retrieval-Augmented Generation) pipelines that ingest mountains of historical mission text, NPRs (NASA Procedural Requirements), and NASA-STDs to map out auditable trace matrices for safety and mission assurance.
Agentic-AI Applications
Autonomous Flight Design & Trajectory Optimizers: Multi-agent swarms that dynamically run Monte Carlo simulations in tools like FreeFlyer or Copernicus, autonomously adjusting parameters to isolate optimal, fuel-efficient orbital trajectories.
Telemetry Anomaly Detonators & Correlators: Multi-agent architectures that sit on top of real-time time-series flight telemetry. When a sensor anomaly triggers, a "diagnostician agent" coordinates with a "historical flight logs agent" to identify the systemic root cause and recommend hardware reconfigurations to the console.
The 600 Group: Professional Administrative
Data-rich, policy-heavy, business-execution environments.
The 600 group manages procurement, legal, institutional security, and Human Capital (OHCM). They leverage enterprise-grade, agentic platforms to streamline bureaucratic processes while maintaining extreme compliance.
Generative-AI Applications
Contract & Proposal Analyzer Engines: GenAI applications that draft and evaluate complex Request for Proposals (RFPs) and Federal Acquisition Regulation (FAR) compliance clauses.
Executive Summary Generators: Transforming multi-hour mission readiness reviews or panel briefings into condensed executive briefs, policy drafts, or public-facing documentation.
Agentic-AI Applications
Autonomous Procurement & Invoice Reconciliation Agents: Multi-agent swarms that manage the supply chain lifecycle. They monitor fulfillment, flag parts delayed by subcontractors, cross-check incoming shipping manifests against baseline contract specs, and handle automated financial matching.
Talent Acquisition Pipeline Orchestrators: Autonomous multi-agent platforms that handle candidate technical validation, parsing resumes against highly specialized AST civil service criteria while managing massive scheduling workflows.
The 500 Group: Clerical & Assistant Support
Procedural, document-centric, high-volume coordination.
The 500 group handles configuration management, document control (such as flight data files and schematics), and center-wide information management.
Generative-AI Applications
Semantic Search & Document Tagging: Custom internal search engines (leveraging geospatial and textual metadata) to immediately locate decades-old engineering drawings, change notices, or Apollo/Shuttle-era structural records.
Smart Correspondence & Templates: Generative tools within secure internal office suites (e.g., FedRAMP-certified Google Workspace / Microsoft 365 Copilot) for drafting formal inter-agency memos, action-item lists, and tracking spreadsheets.
Agentic-AI Applications
Automated Configuration Management (CM) Guards: Agents that monitor document control pipelines. If a design document change is submitted without matching an explicit engineering order or verification signature, the agent autonomously routes notices, flags the discrepancy, and holds the commit.
Smart Meeting & Action-Item Trackers: Enterprise agents that sit passively in non-classified virtual meetings, transcribing, summarizing, and autonomously assigning and following up on action items across different directorates.
The 300 Group: Technical & Nonprofessional Support
Hands-on, physical system integration, testing, and facility operations.
The 300 group operates the machine shops, cleanrooms, and testing facilities (vacuum chambers, acoustics labs, and the Space Vehicle Mockup Facility). Their AI applications bridge digital instructions with physical execution.
Generative-AI Applications
Dynamic Electronic Procedure Generators: GenAI tools that ingest raw engineering specs and generate visual, step-by-step assembly, welding, or inspection manuals tailored for technicians on the cleanroom floor.
Computer Vision Annotation Tools: Generative models used to label training data for vision systems that track automated orbital docking, robot arm grappling, or surface damage inspection.
Agentic-AI Applications
Predictive Facility & Equipment Schedulers: Autonomous agents that monitor the maintenance status, calibration intervals, and usage schedules of heavy machinery, CNC equipment, and cleanroom environmental systems, dynamically adjusting shop schedules when unexpected hardware failures occur.
AI-Augmented Non-Destructive Testing (NDT) Agents: Vision and acoustic-based agents that monitor precision welds or composite layer stackups in real-time, instantly notifying the operator of micro-fractures or structural voids that fall below safety baselines.
Repeat for the Directorate Functional Nodes (Operational Architecture)
Mapping Generative and Agentic AI applications across the Directorate Functional Nodes (Operational Architecture) shifts the perspective from who the employees are (by HR pay grade) to where they sit within the mission and spacecraft lifecycle.
At NASA JSC, the implementation of these technologies relies heavily on Luna, a secure, federally compliant cloud-native platform that orchestrates frontier LLM tokens and sandbox environments, allowing specialized nodes to deploy agentic architectures while strictly isolating flight-critical logic.
1. Center Leadership / Program Offices (Planning & Governance)
Macro-level cross-domain synthesis, program integration, and structural foresight.
Program offices (e.g., ISS, Orion, Gateway) operate under a high data influx from hundreds of subcontractors, international partners, and internal engineering teams. They prioritize macro-level configuration tracking, strategic flight readiness auditing, and policy synthesis.
Generative-AI Applications
Subcontractor Data Harmonization Engines: Generative models that ingest asymmetric technical data formats, configuration baselines, and status reports from distinct commercial vendors (e.g., SpaceX, Boeing, Axiom) and synthesize them into uniform, cross-program progress matrices.
Inter-Agency Policy & FAR Matchers: RAG frameworks that scan evolving federal space policies, FAA/FCC licensing mandates, and Federal Acquisition Regulations (FAR) to flag programmatic friction points for multi-year vehicle roadmaps.
Agentic-AI Applications
Cross-Domain Programmatic Risk Orchestrators: Autonomous agents that query distributed program databases to monitor milestone alignments. If a vendor flags a delayed structural component, the agent traces the dependencies across completely different directorates (e.g., checking if the delay pushes past an FOD training simulator update window or an S&MA certification milestone) and generates predictive impact briefs.
2. Engineering Directorate (EA) (Development)
Deterministic hardware/software prototyping, system engineering, and flight qualification.
The Engineering Directorate designs and certifies spaceflight hardware, avionics architectures, and environmental systems. Their AI applications focus heavily on Model-Based Systems Engineering (MBSE), code generation within standard frameworks like Core Flight System (cFS), and physical optimization.
Generative-AI Applications
Asymmetric Architecture Parsers: GenAI tools that ingest free-form engineering requirements and map them directly into SysML v2 repository blocks, dramatically speeding up early-stage hardware/software interface definition.
Flight Software (FSW) Boilerplate Generators: Sandboxed code LLMs trained exclusively on NASA flight computing parameters (e.g., vxWorks, RTEMS, C/C++) that draft memory-safe, highly predictable code blocks matching strict algorithmic profiles before entering static analysis pipelines.
Agentic-AI Applications
Closed-Loop Multidisciplinary Design Optimization (MDO) Agents: Multi-agent chains tasked with structural or aerodynamic optimization (e.g., Lunar terrain vehicles or descent-stage placement). A "geometry agent" alters a CAD model, a "thermal agent" runs boundary conditions, and a "structural solver agent" calculates stress metrics—cycling iteratively until a design convergence is achieved within physics-based margins.
3. Safety and Mission Assurance (NA) (S&MA)
Independent oversight, rigorous compliance verification, hazard containment, and risk analysis.
The S&MA directorate acts as the ultimate gatekeeper, validating everything against standards like NPR 7150.2D (software engineering) and NASA-STD-8739.8B (software assurance).
Generative-AI Applications
Automated Traceability Matrix Synthesizers: GenAI tools that scan thousands of pages of subsystem requirements, test logs, and verification reports, automatically drafting compliance evidence links to verify that every software line or component maps to an explicitly mandated safety standard.
Hazard Report Augmenters: Models that digest historical anomaly databases (from Apollo to ISS) to assist safety engineers in drafting preliminary hazard analyses, uncovering forgotten failure modes from obsolete systems that share physical or logical architecture with modern projects.
Agentic-AI Applications
The "Holy Grail" Flight Readiness Agent: A highly sophisticated, multi-domain agentic system developed specifically for the Moon to Mars Program.
This engine actively looks across 10 distinct technical and operational data domains to perform continuous risk-based assessments, synthesizing massive, disparate datasets into definitive "Go / No-Go" flight safety recommendations for high-profile milestones (such as the Artemis campaigns). AI-Assisted Software Assurance Case Monitors: Agentic frameworks executing protocols like AMLAS (Assurance of Machine Learning in Autonomous Systems) to continuously monitor the health, data drift, and probabilistic edge cases of AI modules embedded in non-critical systems, keeping assurance cases dynamically updated.
4. Flight Operations Directorate (FOD) (Operations & Training)
Real-time mission tracking, crew readiness execution, and high-fidelity environmental simulation.
FOD manages the execution of human spaceflight, Mission Control Center (MCC) telemetry tracking, the Astronaut Office, and the extensive training infrastructure (including the Space Vehicle Mockup Facility).
Generative-AI Applications
Immersive "Holonaut" Training Specialists: Context-aware, pre-recorded or live AI digital assistants embedded directly within holographic and spatial computing environments (such as HoloNet/HoloConnect architectures). These generative agents interact with training engineers and crew members inside high-fidelity mockups, answering technical design questions or feeding procedural steps in real time.
Multimedia Flight Downlink Annotators: Generative computer vision models that scan downstream video feeds from orbit, auto-populating metadata tags and automatically censoring or flagging streams to protect crew privacy (e.g., catching background proprietary music or tracking astronaut-private living spaces before public logging).
Agentic-AI Applications
Autonomous Mission Operations (AMO) Decision Support Agents: Multi-agent systems operating inside MCC or deep-space habitat analogs designed to counter communication latency (such as lunar or deep-space time delays). These agents sit directly over real-time vehicle telemetry, combining symbolic fault-tree models with probabilistic reasoners (like hybrid diagnosis engines) to isolate complex system anomalies and recommend multi-step recovery scripts directly to controllers or autonomous crew interfaces without waiting for a ground-station handshake.
Crew Timeline & Training Simulator Adjusters: Agents that track individual astronaut physical/cognitive performance metrics across training device software (like the common OnePortal framework) and autonomously adapt future simulation blocks, dynamically injecting specific system failures (ECLSS, GNC) to target identified operational weaknesses.
Repeat for the The 17 Technical Disciplines (Cross-Cutting Technology Taxonomy)
Evaluating the integration of Generative and Agentic AI through the lens of the 17 Technical Disciplines (NASA Technology Taxonomy / TX) transitions the discussion to the explicit R&D, asset management, and intellectual capital tracking utilized across NASA databases like TechPort.
The taxonomy segments the agency's cross-cutting technology baseline.
TX01: Propulsion Systems & TX14: Thermal Management Systems
High-energy physics, computational fluid dynamics (CFD), and thermodynamic boundaries.
Generative-AI Applications
Synthetic Computational Fluid Dynamics (CFD) Accelerators: Generative adversarial networks (GANs) and diffusion models trained on historic hot-fire and cryogenic fluid transfer data to synthesize preliminary injector geometries or nozzle thermal protection layouts, reducing initial supercomputing simulation iterations.
Agentic-AI Applications
Closed-Loop Propellant Management Regulators: Multi-agent architectures that manage automated long-duration cryogenic fuel transfer or deep-throttling lander profiles. If anomalous boil-off or line pressure fluctuations are detected, an agentic loop coordinates thermal valves, pressure regulation, and TVC (Thrust Vector Control) parameters to maintain structural and flight stability.
TX02: Flight Computing and Avionics & TX17: Guidance, Navigation, and Control (GN&C)
Deterministic flight profiles, memory-safe embedded computing, and real-time trajectory optimization.
Generative-AI Applications
Formal Verification Scripting: Generative engines that ingest state machine architectures and automatically output formal mathematical verification scripts (such as Coq or TLA+ baselines) to check for deadlocks or race conditions in Next-Gen Fault-Tolerant Avionics architectures.
Agentic-AI Applications
Autonomous Entry, Descent, and Landing (EDL) Trajectory Replanners: Multi-agent loops embedded at the chip level that fuse real-time hazard detection camera maps with GNC loops. If a crater or boulder is identified during landing, the agents collectively negotiate thruster firings and attitude estimation adjustments in milliseconds to recalculate optimal terrain-relative touchdown paths.
TX03: Aerospace Power and Energy Storage
Stochastic chemical/solar deterioration tracking, microgrids, and load optimization.
Generative-AI Applications
Battery Degradation Profiles: Generative models that ingest asymmetric electrochemical data to simulate and generate synthetic aging profiles for Lithium-Sulfur or solid-state batteries exposed to cyclic extreme thermal swings (e.g., lunar night).
Agentic-AI Applications
Autonomous Planetary Microgrid Balancing Agents: Multi-agent networks operating over surface habitats or rovers. A "generation agent" tracks solar/nuclear source status, a "storage agent" monitors state-of-charge, and "load agents" prioritize life support (ECLSS) over science payloads during unexpected dust storms, dynamically shedding loads to protect core vehicle health.
TX04: Robotic Systems & TX10: Autonomous Systems
Kinematics, human-machine teaming, situational awareness, and reasoning/acting paths.
Generative-AI Applications
Reinforcement Learning (RL) World Model Simulators: Generative frameworks (such as world models derived from generative neural architectures) that synthesize 60-second interactive loops of planetary surfaces using orbital imagery, allowing robotic rovers to train on visual and physical terrain properties before moving.
Agentic-AI Applications
Dynamic Multi-Agent Swarm Coordinators: Autonomous agentic networks that orchestrate teams of heterogeneous surface robots (e.g., excavation rovers and micro-scouts). The agents handle spatial task-sharing, negotiate path routes over treacherous terrain, and self-heal communication mesh configurations without requiring human intervention.
TX05: Communications, Navigation, and Orbital Debris Tracking
RF/Optical spectrum allocation, Delay-Tolerant Networking (DTN), and space domain tracking.
Generative-AI Applications
Cognitive Radio Spectrum Shapers: Generative algorithms that predict and synthesize real-time signal modulation patterns to overcome unexpected solar radio bursts or ionospheric scintillation during deep space communications.
Agentic-AI Applications
Autonomous Conjunction Assessment & Collision Avoidance Loops: Orbital debris tracking agents that parse active tracking catalogs, calculate real-time probability of collision ($P_c$), and autonomously calculate, test, and command evasive delta-V maneuvers to orbiting assets.
TX06: Human Health, Life Support, and Habitation Systems (ECLSS)
Environmental closed-loops, metabolic tracking, and Human Systems Integration (HSI).
Generative-AI Applications
Biomedical Trend Synthesizers: GenAI tools that ingest telemetry from wearables and life-support systems to generate synthetic longitudinal astronaut health trends, identifying sub-clinical physiological degradation (such as bone density or ocular shift markers).
Agentic-AI Applications
Predictive ECLSS Anomaly Countermeasure Loops: An agentic framework monitoring closed-loop carbon dioxide removal and water reclamation assemblies. If a trace contaminant spikes, a diagnostic agent isolates the failing bed or filtration matrix, coordinates with logistics to verify replacement part availability, and generates a procedural manifest for crew maintenance.
TX07: Exploration Destination Systems & TX13: Ground, Test, and Surface Systems
ISRU (In-Situ Resource Utilization), surface infrastructure, site preparation, and launch operations.
Generative-AI Applications
Generative Lunar Base/Structure Synthesizers: Topology optimization models that ingest native regolith material constraints, structural stress limits, and radiation shielding metrics to generate structurally optimized 3D-printable habitats or protective berm designs.
Agentic-AI Applications
Autonomous Regolith Processing & Task Schedulers: Agentic loops controlling automated mining, sorting, and chemical processing facilities for ISRU liquid oxygen or water production, dynamically re-routing resources and power profiles as feedstock density changes.
TX08: Sensors and Instruments & TX09: Entry, Descent, and Landing
Remote sensing, hyper-spectral mapping, and aeroassist atmospheric modeling.
Generative-AI Applications
Hyper-Spectral Super-Resolution Synthesizers: Generative networks that transform coarse satellite or instrument readings into super-resolved imagery by inferring localized fine structural details based on trained multi-instrument datasets.
Agentic-AI Applications
Self-Calibrating Deep Space Instrument Arrays: Autonomous agent networks embedded within scientific instrument packages (e.g., spectrometers or magnetometers) that track internal sensor drift, isolate background noise or magnetic signatures of the host spacecraft, and execute micro-adjustments to maintain calibration parameters during long-duration flight.
TX11: Software, Modeling, Simulation, and Information Processing
Software integrity, high-performance computing (HPC), and Artificial Intelligence development.
This discipline functions as the core engine for NASA's modern AI strategy—utilizing machine learning mechanisms (like the internal T-Rex Taxonomy Recommender API) to classify new technical research while managing the lifecycle of AI itself.
Generative-AI Applications
Automated Assurance Case Architects: Generative tools that draft dynamic software assurance documentation by mapping raw Git commits and continuous integration (CI) logs directly to standards like NPR 7150.2D.
Agentic-AI Applications
Federated Model Drift Monitors: Agentic nodes running across local and distributed computing clusters that actively evaluate the precision, safety envelopes, and behavioral shifts of other embedded AI models, triggering automated sandboxed re-training pipelines if a model's performance begins to degrade or experience unexpected epistemic uncertainty.
TX12: Materials, Structures, Mechanical Systems, and Manufacturing & TX15: Flight Vehicle Systems
Metals, composites, additive manufacturing, structural dynamics, and vehicle lifecycle systems.
Generative-AI Applications
Inverse Material Design Networks: Generative models that explore unexplored crystal lattice structures or composite orientations based on specified performance inputs (e.g., "Generate a crystalline alloy layout that yields a 15% increase in tensile strength at $1200^{\circ}\text{C}$").
Agentic-AI Applications
In-Situ Additive Manufacturing Quality Control Guards: Vision-based and ultrasonic monitoring agents integrated into large-format metal 3D printers. The agents evaluate layer-by-layer material deposition in real-time; if a microscopic void or thermal warping anomaly is detected, the agent modifies the printing laser path or powder deposition parameters mid-print to preserve structural integrity.
TX16: Air Traffic Management and Range Tracking Systems
Safe airspace integration, commercial space lanes, and range surveillance.
Generative-AI Applications
Dynamic Airspace Restructuring Simulators: Generative applications that output synthetic, chaotic traffic and severe weather combinations to evaluate how commercial space launch corridors impact normal civil aviation routes.
Agentic-AI Applications
Autonomous Range Flight Safety Evaluators: Real-time multi-agent systems that coordinate range tracking data, tracking launch vehicle tracking anomalies against instantaneous impact prediction footprints to deliver rapid flight termination recommendations or automated corridor clearance updates during active launches.
From <https://gemini.google.com/app/03b6eb2a82ce279d> Google Gemini (3.5 Flash)
Comments
Post a Comment