About Us • Draccon Tech

Our Mission

Transform stellar signals into actionable knowledge in near real time.
Bridge data science, engineering, and design for responsible exploration.
Reduce latency through autonomous edge computation and intelligent mediation.

How It Works

Detection (AION): Scans systems, identifies candidates, and builds dossiers.
Protection (EXON): Validates packets, filters anomalies, and self-corrects errors.
Routing (NOVA): Decides what stays at the edge, compresses data, and optimizes communication paths.
Consensus (EIRYA): Consolidates knowledge, version-controls models, and replicates the swarm’s neural network.

Data Source & API

This prototype consumes public exoplanet parameters from NASA’s Exoplanet Archive. We use the TAP endpoint below (JSON), which can be queried directly:

https://exoplanetarchive.ipac.caltech.edu/TAP/sync?query=select+pl_name,pl_rade,pl_masse,pl_eqt+from+pscomppars&format=json

Example (curl)

curl "https://exoplanetarchive.ipac.caltech.edu/TAP/sync?query=select+pl_name,pl_rade,pl_masse,pl_eqt+from+pscomppars&format=json"

On-Orbit Architecture: Why Four AIs (Not One)

In deep-space operations, latency, intermittent links, and radiation-induced faults make ground-only processing impractical. Placing the intelligence on-orbit (inside satellites and edge probes) enables real-time decisions under tight downlink budgets. SWARM is intentionally split into four specialized AIs to maximize fault isolation, throughput, and mission assurance.

AION — On-Board Science

Runs on the instrument bus to detect, rank, and characterize candidates in situ. Performs lightweight spectral filtering, transit-curve spotting, and feature extraction close to the sensor, reducing raw data volume before it ever leaves the spacecraft.

Minimizes downlink by sending dossiers instead of raw streams.
Adapts thresholds to detector noise/temperature drift.

EXON — Security & Self-Heal

The security and reliability layer. Checks packet integrity, monitors SEUs (single-event upsets), and performs error correction/model rollback. Partitions faults to prevent common-mode failures.

Lightweight crypto and integrity checks on board.
Health monitoring and model/data self-repair.

NOVA — Mediation & Routing

Orchestrates traffic between nodes and ground. Decides what stays at the edge, what to compress, and when to transmit based on link windows and available power. Keeps the mesh alive with DTN.

DTN-aware scheduling, compression, and store-and-forward.
QoS by science priority and constellation state.

EIRYA — HiveMind Backbone

Maintains the swarm’s global state: versions models, distributes weights, consolidates consensus across nodes, and synchronizes knowledge when links allow. The fabric that keeps everything coherent.

Eventual consensus and parameter replication.
Mission-safe model rollout/rollback policies.

Why Not a Single Monolithic AI?

Fault Isolation: Domain specialization prevents a science fault from taking down security/routing.
Operational determinism: Critical paths (telemetry, security) remain predictable and auditable.
Real-time at the edge: AION decides at the sensor; NOVA optimizes link windows; no waiting for ground.
Safe updates: EIRYA rolls versions in waves EXON reverts on regression.
Hardware efficiency: Pipelines map cleanly to distinct compute tiles (CPU/GPU/FPGA/rad-hard cores).
Scalability: Easier to add/remove nodes than to retrain a critical monolith.

In short:four cooperating AIs deliver resilience, low latency and operability versus a single monolithic agent—critical in radiation-rich, power-limited, and window-scarce environments.