Richard Cloete

Overview

Richard Cloete is a computer scientist and Laukien-Oumuamua Postdoctoral Research Fellow at the Center for Astrophysics | Harvard & Smithsonian. He is a member of the Galileo Project and has co-authored technical work on computing infrastructure, multimodal observatories, infrared imaging and automated classification for the scientific study of aerial phenomena. [S1][S2]

In June 2026, Cloete was publicly named to the UAP Science Advisory Council (UAPSAC), with data analysis and artificial-intelligence tools identified as his role. Independent reporting by DefenseScoop confirmed that UAPSAC is one of the external advisory groups supporting the ODNI-linked UAP Governance Board. [S3][S4]

Cloete is one of the strongest UAPSAC additions for UAPRAD because his UAP-specific work predates the council and is technically inspectable.

His contribution concerns a foundational problem: how can a continuous sensor network process enormous numbers of ordinary aerial events, identify known classes reliably and isolate statistically unusual observations without confusing a machine-learning outlier with a physically anomalous object?

That distinction—outlier versus anomaly—is central to evaluating AI in UAP research.

Computer-science background

The Center for Astrophysics identifies Cloete as a computer scientist with a PhD in computer science. The Galileo Project states that he previously held a postdoctoral position at the University of Cambridge before joining Harvard. [S1][S5]

Newcastle University's doctoral record provides additional detail.

Cloete's PhD research concerned latency measurement, modelling and management in interactive remote rendering, including prediction, simulation and performance measurement in cloud-based systems. [S6]

This background is relevant to the data-engineering side of UAP research.

A modern multimodal observatory does not merely collect photographs.

It has to:

• synchronise multiple sensors;
• stream large data volumes;
• detect objects in real time;
• track them across frames;
• compare observations with external catalogues;
• classify known objects;
• preserve raw data;
• flag candidate outliers for later analysis.

These are computing and systems-engineering problems before they become questions about exotic physics.

Galileo Project role

The Galileo Project's current profile describes Cloete as a research member who defines and conducts project research under Avi Loeb at the Center for Astrophysics. [S1]

The Project seeks to replace anecdotal UAP evidence with systematic measurements from ground-based observatories.

Its technical architecture includes optical, infrared, radar, radio-frequency, acoustic and environmental sensors.

The computing layer is essential because a monitoring station may observe enormous numbers of aircraft, satellites, birds, insects, weather events and astronomical sources.

Most detections should be ordinary.

A scientifically credible system therefore has to become very good at recognising the ordinary before an unusual residual becomes informative.

2023 integrated computing platform paper

Cloete was lead author of the 2023 paper "Integrated Computing Platform for Detection and Tracking of Unidentified Aerial Phenomena (UAP)" in the Journal of Astronomical Instrumentation. [S2]

The paper describes a proposed computing architecture for ingesting data from multiple sensor types, processing observations in real time and distinguishing natural and human-made phenomena from events that fall outside the known phenomenological envelope.

The design includes algorithms for offline analysis as well as real-time detection.

The paper explicitly describes the work as preliminary.

This matters.

A proposed architecture is not evidence that the system has already detected a novel UAP.

Its scientific contribution is engineering: defining how data should move from sensors to classification and outlier analysis.

Outliers versus scientific anomalies

The integrated-platform paper makes an important conceptual distinction between statistical outliers and scientific anomalies. [S2]

A statistical outlier is a measurement that differs from the dominant distribution.

That can happen for many reasons:

• bad calibration;
• rare but known aircraft;
• unusual weather;
• data corruption;
• sensor saturation;
• an insect very near a camera;
• incomplete classification data.

A scientific anomaly is a much stronger category.

It requires a corroborated and statistically distinctive observation that resists explanation under prevailing models after artefacts and known alternatives are tested.

This distinction is essential because AI systems are exceptionally good at producing "weird" candidates.

A high anomaly score is not evidence of non-human technology.

It is a triage signal telling researchers where to look more closely.

Multimodal observatory design

Cloete also co-authored the 2023 Galileo Project roadmap "The Scientific Investigation of Unidentified Aerial Phenomena (UAP) Using Multimodal Ground-Based Observatories." [S7]

The paper describes a science traceability matrix linking desired physical parameters to observable quantities and instrument requirements.

The sensor package includes:

• wide-field cameras;
• narrow-field multispectral cameras;
• passive multistatic radar;
• radio-spectrum monitoring;
• acoustic sensors;
• environmental measurements.

The objective is not simply to photograph something unusual.

It is to recover enough independent measurements to estimate physical properties such as position, trajectory, spectrum and kinematics and to determine whether apparent anomalies survive cross-sensor comparison.

This approach directly addresses weaknesses in many historical UAP cases.

Aerial object localisation

Cloete co-authored a separate 2023 paper describing hardware and software for three-dimensional aerial-object localisation.

The system uses multiple cameras across visible, near-infrared and infrared bands, calibration and triangulation to recover positions over time and thereby estimate velocity and acceleration. [S8]

This is especially relevant to modern UAP video debates.

A single image sequence can produce apparent rapid motion without providing reliable range.

A multi-camera calibrated system can recover depth geometrically.

That turns angular motion into measurable kinematics.

The engineering challenge is therefore not merely higher resolution.

It is acquisition of independent geometry.

Acoustic and radar systems

Cloete is also listed as a co-author on Galileo Project work concerning multi-band acoustic monitoring and passive multistatic radar. [S9][S10]

The acoustic system is designed to monitor infrasonic, audible and ultrasonic signatures and use known aircraft recordings to develop identification algorithms.

The radar concept is intended to recover object position and velocity from distributed receivers.

These modalities are scientifically valuable because they can corroborate or contradict optical data.

An optical target with no corresponding acoustic, radar or thermal signature may still be real, depending on range and sensor sensitivity, but the multisensor pattern provides constraints that a video alone cannot.

2025 infrared camera commissioning

In 2025, Cloete co-authored a peer-reviewed Sensors paper on commissioning an all-sky infrared camera array for airborne-object detection. His listed contributions included software, formal analysis and original drafting. [S11]

Commissioning work is important because scientific instruments must be characterised before their anomalies can be trusted.

Researchers need to know:

• sensitivity;
• dead pixels;
• lens distortion;
• thermal behaviour;
• background noise;
• false detections;
• image-processing effects.

An uncalibrated sensor can create convincing anomalies that disappear once detector behaviour is understood.

Cloete's involvement in commissioning therefore strengthens his relevance to UAPSAC's data-analysis role.

Machine learning outside UAP

Cloete's current research is not confined to UAP.

In 2024, he co-authored a peer-reviewed Astronomy & Astrophysics paper on machine-learning methods for automated classification of interstellar objects in the Vera C. Rubin Observatory/LSST data environment. [S12]

In 2025, he co-authored another Astronomy & Astrophysics paper applying machine-learning methods to improve discovery of near-Earth objects. [S13]

These publications are important to his profile because they demonstrate machine-learning work in established astronomical classification problems where ground truth and survey performance can be tested.

The relevance to UAP is methodological.

Algorithms should first prove that they can identify known populations, estimate error rates and generalise to new data before researchers place strong weight on their classification of an "unknown."

AI and UAP

Artificial intelligence is attractive in UAP research because sensor networks create too much data for manual review.

But AI introduces its own epistemological risks.

A model can learn artefacts in the training data.

It can classify an object with high confidence even when presented with a category absent from training.

It can produce an "unknown" simply because metadata are missing.

Out-of-distribution detection is itself an active research problem.

Accordingly, a UAP AI pipeline needs more than sophisticated software.

It needs:

• known-object validation sets;
• human review;
• transparent thresholds;
• reproducible preprocessing;
• model-version records;
• uncertainty calibration;
• access to raw sensor data.

Cloete's technical role is therefore potentially central to whether UAPSAC avoids false positives.

Joining UAPSAC

Cloete appeared in the initial June 2026 council roster with responsibility for data analysis and AI tools. [S3][S4]

Unlike members whose contributions are primarily policy or public communication, his assigned role is close to the core technical question of turning UAP data into classifications.

This does not mean the council's AI systems have already produced a discovery.

Publicly available information reviewed here does not provide a complete UAPSAC data pipeline, model architecture or validated case-analysis output.

His profile should therefore describe the appointment as a new application of an established research speciality rather than evidence of a completed council result.

Relationship to Avi Loeb and independence

Cloete works within the Galileo Project under Avi Loeb and is now on a council chaired by Loeb.

This creates both continuity and a potential methodological concern.

Continuity is useful because existing tools and expertise can be transferred rapidly into government advisory work.

The concern is independence.

If the same research group designs instrumentation, selects candidate anomalies and evaluates them without sufficient external replication, confirmation bias can enter even in technically sophisticated systems.

The safeguard is not to exclude affiliated researchers.

It is to ensure independent validation, open methods where possible and clear separation between algorithm development and final anomaly claims.

Evidence analysis

Cloete's profile has a strong documentary and scientific foundation.

His Harvard role, PhD background and Galileo Project work are directly documented.

His UAP technical publications are peer-reviewed and describe concrete computing and instrumentation systems.

The most important limitation is maturity.

Much of the Galileo Project literature describes architecture, commissioning and capability rather than a confirmed novel phenomenon.

This is not a weakness of the research programme.

It simply means the evidential contribution is currently methodological.

The future test is whether the system can classify known objects at high accuracy and then produce candidate events with enough multisensor data for independent physical analysis.

What is established

• Cloete is a computer scientist and postdoctoral research fellow at the Center for Astrophysics | Harvard & Smithsonian. [S1][S5]
• He holds a PhD in computer science from Newcastle University. [S6]
• He is a Galileo Project research member. [S1]
• He led the 2023 integrated computing-platform paper for UAP detection and tracking. [S2]
• He co-authored Galileo Project work on multimodal observatories, localisation, acoustic monitoring and radar. [S7][S8][S9][S10]
• He co-authored a 2025 peer-reviewed all-sky infrared commissioning paper. [S11]
• He has mainstream astronomy machine-learning publications on interstellar and near-Earth object classification. [S12][S13]
• He is a UAPSAC member assigned to data analysis and AI tools. [S3][S4]

What is not established

• An AI-flagged outlier is not automatically a scientific anomaly.
• A scientific anomaly is not automatically evidence of non-human technology.
• Galileo Project technical papers do not establish that its observatories have detected extraordinary craft.
• Machine-learning expertise does not eliminate training bias, classification error or out-of-distribution problems.
• UAPSAC membership does not establish that Cloete has analysed decisive classified UAP evidence.
• Affiliation with Harvard or the Center for Astrophysics does not make every Galileo/UAPSAC interpretation an institutional Harvard finding.

Missing or unavailable evidence

The principal missing evidence concerns performance at scale.

To assess a mature UAP AI system, researchers would need published information on:

• training and validation datasets;
• precision and recall by object class;
• false-positive rates;
• anomaly-score calibration;
• cross-sensor fusion performance;
• handling of unknown classes;
• human-review procedures.

For UAPSAC specifically, publication of model and pipeline documentation would make it possible to assess whether government UAP data are being analysed with reproducible standards.

Overall assessment

Richard Cloete is one of the strongest technically grounded additions to the UAPRAD Humans index.

His importance does not depend on extraordinary claims.

It comes from building the computational infrastructure required to decide whether an observation is extraordinary in the first place.

The Galileo Project's approach recognises a basic reality: a sky-monitoring system will detect enormous numbers of ordinary objects.

The scientific problem is therefore dominated by classification, calibration and false-positive rejection.

Cloete's work on integrated pipelines, multimodal sensors and machine learning directly addresses that bottleneck.

His mainstream astronomical machine-learning work strengthens the case for inclusion because it demonstrates similar methods in domains where classification quality can be benchmarked.

The principal caution is semantic.

"Outlier", "unidentified" and "anomalous" are not interchangeable.

An AI model can identify a statistical outlier for trivial reasons.

Only after artefacts, known objects and model limitations are excluded does a candidate become scientifically interesting.

For UAPRAD, Cloete's profile can help readers understand that AI does not solve UAP by labelling strange images.

Its real value is making the ordinary sky sufficiently well characterised that a genuinely unusual residual, if one appears, can be recognised and measured.

Confidence by proposition

Sources

[S1] Primary institutional source — The Galileo Project. Richard Cloete profile. https://galileo.hsites.harvard.edu/people/richard-cloete

[S2] Peer-reviewed research — Richard Cloete et al. "Integrated Computing Platform for Detection and Tracking of Unidentified Aerial Phenomena (UAP)," Journal of Astronomical Instrumentation 12 (2023). https://galileo.hsites.harvard.edu/publications/integrated-computing-platform-detection-and-tracking-unidentified-aerial

[S3] Primary council-roster source — Disclosure Foundation. UAP Science Advisory Council, June 2026. https://disclosure.org/news/uap-science-advisory-council

[S4] Independent government-sourced reporting — DefenseScoop. UAPSAC/Governance Board, 17 June 2026. https://defensescoop.com/2026/06/17/new-science-advisory-council-forms-to-help-us-government-resolve-the-uap-mystery/

[S5] Primary institutional source — Center for Astrophysics | Harvard & Smithsonian. Richard Cloete. https://www.cfa.harvard.edu/people/richard-cloete

[S6] Primary academic record — Newcastle University. Richard Cloete, PhD thesis, Latency Measurement, Modelling and Management for Interactive Remote Rendering (2020). https://theses.ncl.ac.uk/jspui/handle/10443/5058

[S7] Peer-reviewed research — Wesley A. Watters et al. "The Scientific Investigation of Unidentified Aerial Phenomena (UAP) Using Multimodal Ground-Based Observatories," Journal of Astronomical Instrumentation 12, 2340006 (2023). https://galileo.hsites.harvard.edu/publications/scientific-investigation-unidentified-aerial-phenomena-uap-using-multimodal

[S8] Peer-reviewed research — Matthew Szenher et al. "A Hardware and Software Platform for Aerial Object Localization," Journal of Astronomical Instrumentation 12, 2340002 (2023). https://galileo.hsites.harvard.edu/publications/hardware-and-software-platform-aerial-object-localization

[S9] Peer-reviewed research — Andrew Mead et al. "Multi-Band Acoustic Monitoring of Aerial Signatures," Journal of Astronomical Instrumentation 12, 2340005 (2023). https://galileo.hsites.harvard.edu/publications/multi-band-acoustic-monitoring-aerial-signatures

[S10] Peer-reviewed research — Mitch Randall et al. "SkyWatch: A Passive Multistatic Radar Network for the Measurement of Object Position and Velocity," Journal of Astronomical Instrumentation (2023). https://galileo.hsites.harvard.edu/publications/skywatch-passive-multistatic-radar-network-measurement-object-position-and

[S11] Peer-reviewed research — Laura Domine et al. "Commissioning an All-Sky Infrared Camera Array for Detection of Airborne Objects," Sensors 25(3):783 (2025). https://pmc.ncbi.nlm.nih.gov/articles/PMC11820869/

[S12] Peer-reviewed mainstream astronomy — Richard Cloete, Peter Vereš and Abraham Loeb. "Machine learning methods for automated interstellar object classification with LSST," Astronomy & Astrophysics 691, A338 (2024). https://www.aanda.org/articles/aa/pdf/2024/11/aa51118-24.pdf

[S13] Peer-reviewed mainstream astronomy — Peter Vereš, Richard Cloete, Matthew J. Payne and Abraham Loeb. "Improving the discovery of near-Earth objects with machine-learning methods," Astronomy & Astrophysics 698, A242 (2025). https://doi.org/10.1051/0004-6361/202554311