OpticalX, LLC /Tracy, California /Est. 2022

A new definition
of to see.

Computational optics for the objects conventional sensors cannot detect — too small for radar, too faint for classical telescopes, too fast for either. We build the arrays and the processing that make them visible — fielded today, delivering thousands of tracking records to the Unified Data Library nightly.

See the instrument Capabilities
Continuous operations
16months
Observation records
1M+
Objects tracked
21,000RSOs
Delivered to UDL
2–4kobs/day
01 The Instrument

Signal, recovered
from noise.

In any single short exposure, a five-centimeter object in low Earth orbit sits beneath the noise floor. It is not dim. It is invisible.

Synthetic tracking recovers it. Hundreds of frames are shifted and added along a hypothesized velocity vector: signal accumulates linearly, noise only as its square root. The object climbs out of the background as √N — and the fixed stars, no longer registered, smear into streaks.

SYNTHETIC TRACKING · SHIFT-AND-ADD ALONG VELOCITY HYPOTHESIS
v⃗ = [ +1.35, −0.42 ] px · frame−1
DETECTION · 5σ THRESHOLD CROSSED
Frames stacked 1
Integration 5ms
Detection SNR 0.42σ
Status BELOW NOISE FLOOR
Interactive simulation. Drag to accumulate frames — the object crosses the 5σ detection threshold at roughly 140 frames, less than a second of integration.

The same principle finds near-Earth asteroids for planetary defense. Applying it in low Earth orbit is harder by orders of magnitude: objects cross the field of view in a fraction of a second, demanding roughly a hundred times the frame rate and petabyte-scale analysis every night. That computation is the product — and it is measured, not modeled: 10–50× SNR gains over conventional imaging, with the radiometric framework peer-reviewed in Optics Express (2026).

02 Capabilities

Two problems,
one instrument.

Space Domain Awareness

Persistent custody
of the untrackable.

An estimated one million objects between 1 and 10 cm are in orbit. Well under one percent are cataloged. Each carries enough kinetic energy to end a mission, and none of them appear on a screen before they arrive.

This is fielded capability, not a roadmap. Our SPOT-Imager optical fence and 0.2–1.0 m telescope network — sites across California and New Mexico, with partner apertures in Chile — has operated continuously for sixteen months, processing on GPU edge nodes and delivering calibrated astrometric, photometric, and spectroscopic products to the Unified Data Library. Detections resolve into orbits; orbits resolve into conjunction warnings hours to days ahead.

  • 10 cm objects tracked at 1,000 km — invisible in any single frame
  • Validated across LEO, MEO, GEO, and HEO regimes
  • Edge GPU processing — no raw-data backhaul
  • Live UDL delivery: 2,000–4,000 tracking records daily
  • Capacity scales by adding nodes, not by scaling aperture

Event Characterization & Assessment

Faster, higher-confidence
assessment.

A breakup, a shed component, an anomalous maneuver — an on-orbit event announces itself first as a change in the debris field around it. Resolving the fragments is what turns an unexplained observation into a characterized event.

These analytics are demonstrated, not proposed — delivered as CCDM services during a nine-month U.S. Space Force SDA TAP Lab engagement (Cohorts 7–9). A year-deep baseline of a million light-curve records grounds every call: anomaly is measured against an object’s own pattern of life, with quantified confidence attached, in formats operational systems already consume.

  • Light-curve anomaly detection: tumble onset, signature change
  • Breakup, separation, and sub-satellite deployment detection
  • Rapid maneuver detection on the optical fence
  • HAMR characterization of shed debris — foil, panel fragments
  • Multi-source fusion: optical, radar, and catalog priors
03 Pipeline

From photon
to product.

One continuous path, automated end to end. Every stage is instrumented, every output carries its provenance, and the latency budget is measured against the decision it has to support — not against the convenience of the processing.

  1. 01

    Collect

    Distributed telescope nodes at high frame cadence. Commercial optics, hardened scheduling, autonomous nightly operation.

    102–3 fps · PB / night
  2. 02

    Process

    GPU shift-and-add across the velocity hypothesis space. Petabytes reduced to candidate detections at the edge, before any link is crossed.

    Synthetic tracking · on-node
  3. 03

    Exploit

    Orbit determination, correlation against catalog, and fusion with radar and external sources. Events characterized, confidence quantified.

    Fusion · OD · confidence
  4. 04

    Disseminate

    Assessments and warnings delivered in standard formats to the systems that already carry them, with full provenance back to the originating frames.

    Standards-based · traceable
04 Services

What we
deliver.

Six service lines, engaged individually or as a whole. We work as prime on focused efforts and as a subcontractor where our processing plugs into a larger system — the algorithms and the pipeline travel either way.

S1

Computational optics R&D

Synthetic tracking and shift-and-add across velocity hypothesis space. Detection of objects that sit below the single-frame noise floor, where conventional thresholding returns nothing.

S2

Sensor node design & deployment

Telescope arrays assembled from commercial optics — specified, integrated, sited, and left to run autonomously through the night without an operator on console.

S3

GPU processing pipelines

High-cadence data reduction on parallel computing platforms. Petabyte-scale nightly throughput reduced to candidate detections at the edge, before any link is crossed.

S4

Orbit determination & custody

Detection-to-orbit processing, correlation against catalog, and conjunction screening — maintaining custody of objects that were never cataloged to begin with.

S5

Fusion & event assessment

Optical detections fused with radar tracks and catalog priors, resolved into characterized events with quantified confidence and full provenance.

S6

Transition engineering

Maturation from validated prototype toward fielded system: containerized delivery, commercial supply chain, and integration with platforms already in service.

05 Transition

Built to
transition.

Capability that stays in a laboratory is not capability. OpticalX designs for the handover from the outset: commercially available optics, containerized GPU software, and turn-key deployment into environments that already exist.

Moving a validated prototype into a fielded, accredited system is an engineering problem in its own right. We treat it as one — alongside the physics, not after it.

Technology readiness · SPOT-Imager network Fielded · TRL 7 → 8
  1. TRL 4Component validation in laboratory
  2. TRL 5Validation in relevant environment
  3. TRL 6Prototype demonstration
  4. TRL 7Operating against the live UDL, 16 months continuous
  5. TRL 8System qualified, ready to field

Commercial hardware

Off-the-shelf apertures and sensors. No exquisite optics, no bespoke supply chain, no single-vendor dependency.

Containerized software

The processing chain deploys as containers onto GPU hardware — and the fielded edge nodes already host third-party algorithms in production.

Turn-key deployment

Nodes ship configured and operate autonomously through the night, reporting products rather than raw data.

Additive by design

Coverage, revisit rate, and sensitivity all improve by adding nodes to a network that is already running.

06 Past Performance

Work delivered
under contract.

Federally funded research delivered as prime — NSF and U.S. Air Force SBIR awards, an active Missile Defense Agency vehicle, and a nine-month U.S. Space Force SDA TAP Lab engagement. Additional references and detailed technical reporting are available on request.

National Science Foundation · America’s Seed Fund

Space-Time Projection Optical Tomography (SPOT)

Complete
Role
Prime
Program
SBIR Phase I
Award no.
2404362
Value
$274,996
Period
Jul 2024 – Dec 2025
Place
Tracy, CA

Scope

Development of an optical solution for space debris detection using a small array of telescopes and algorithms implemented on GPU-based parallel computing platforms — targeting the population in low Earth orbit that current radar and optical techniques cannot detect, and establishing the synthetic-tracking chain that underpins every service line above.

Measured results

Real-time GPU pipeline fielded on Jetson-class edge nodes; 10 cm objects detected and tracked at 1,000 km — invisible in single frames; 10–50× SNR gain over conventional imaging; validation across LEO, MEO, GEO, and HEO; live UDL integration at 2,000–4,000 records daily; radiometric framework published in Optics Express (2026).

Phase-Space Imager

Applied research under the U.S. Air Force SBIR program — extending phase-space methods for the imaging chain.

USAF SBIR Phase I · AFRL/AFWERX
FA8649-24-P-0792 · $74,792 · May–Aug 2024
Complete

SDA TAP Lab · Cohorts 7–9

CCDM services to the Welders Arc federation: light-curve stability-change detection, breakup and separation detection, rapid maneuver detection on the optical fence.

U.S. Space Force · Colorado Springs, CO
Nine-month engagement · 2025–2026
Complete

SHIELD

Scalable Homeland Innovative Enterprise Layered Defense — awardee on the multiple-award vehicle, eligible to compete for task orders through 2035.

Missile Defense Agency
HQ0859-26-F-F745 · ordering period to Dec 2035
Active

Who we are

Founded in 2022, OpticalX is a small team of engineers and consultants with backgrounds in remote sensing, data science, parallel computing, and space physics. Our mission is to develop effective computational optics solutions with turn-key deployments across space domain awareness, debris monitoring, and aircraft surveillance.

Leadership

Hasan Bahcivan

Co-founder & Chief Scientist · Principal Investigator

Previously with the Center for Geospace Studies at SRI International. Principal investigator on ground-to-space bistatic radar experiments flown aboard the Radio Aurora Explorer CubeSats, with twenty years of radar phenomenology across NSF’s AMISR, Arecibo, and Sondrestrom facilities and published research spanning radar and optical remote sensing.

Collaboration

We would like to work
on the hard ones.

Partnerships, pilots, and prime or subcontract teaming across space domain awareness, debris custody, and on-orbit event assessment.

hasan.bahcivan@opticalx.space