2026 Cohort
Project Portfolio

The severe physical constraints of orbital downlink bandwidth significantly limit Earth Observation satellites, causing sensors that capture vast amounts of daily data to transmit only a small fraction during brief ground station passes. Because current bent-pipe architectures blindly transmit raw files in a first-in, first-out manner, significant volumes of potentially critical intelligence are overwritten and permanently lost per satellite daily. This bottleneck forces operators to lose contract-grade intelligence, meaning vital data regarding disaster responses, defense tasking, and commercial analytics can expire on orbit before reaching Earth. To overcome this physical barrier, the project proposes a value-aware onboard triage system that evaluates data directly on the satellite to filter out low-priority regions, extract semantic vector coordinates, and rank transmissions by urgency. By shifting the paradigm from transmitting massive raw pixel files to delivering lightweight, high-value intelligence, operators can maximize the utility of every downlinked byte and fundamentally change their business models to provide real-time information.

The potential under-monetization of Earth Observation tasking capacity faced by large commercial operators is driven by current business models that heavily favor subscriptions, archive sales, and custom enterprise contracts over the rapid allocation of short-notice availability. Beyond simply leaving eligible collection windows unpriced, this reliance on contract-centric access risks lower overall revenue for operators investing heavily in taskable constellation infrastructure. The project explores how to bridge this gap by focusing strictly on operator-side utilization, pricing, and risk management rather than end-user access, targeting the lack of systems designed for the dynamic, low-friction allocation of unallocated windows. By treating third-party estimates of high unused capacity as hypotheses to be validated through capacity logs and operator interviews, the initiative aims to develop strategies for rapid capacity allocation and better yield management for space infrastructure operators.

The growing wildfire risk faced by rural communities is being intensified by climate change, land abandonment, fragmented land management, limited prevention resources and the accumulation of unmanaged forest biomass, making fires increasingly frequent and destructive. Beyond their environmental impact, wildfires can devastate local economies by destroying pastures, livestock, agricultural infrastructure and entire livelihoods, particularly in regions that depend on the primary sector. The project explores how data-driven tools can be combined with traditional land-management practices to turn forest biomass into a valuable resource, supporting cleaner landscapes, earlier risk identification, stronger prevention strategies and more informed decision-making for stakeholders working on land management, environmental protection and rural resilience.

Large satellite operators face a critical challenge when a severe anomaly or terminal failure occurs: the diagnostic evidence needed to understand the root cause can be lost together with the spacecraft. In complex missions involving high-value assets, shared subsystems and long operational lifetimes, failures in the power system, onboard computer or telemetry chain can trap, overwrite or destroy the very data engineers need to reconstruct what happened. Current spacecraft architectures are primarily designed to keep missions alive and recover from faults, but they are not optimised to preserve the final diagnostic state for post-failure analysis. This project addresses that gap by exploring an independent evidence-preservation layer capable of safeguarding key telemetry during critical events, helping operators turn unresolved mission losses into actionable engineering lessons for future spacecraft.

Mechanical engineers at early-stage aerospace hardware teams (5 to 50 people, venture-funded, no dedicated PLM or systems engineer) hit avoidable redesign cycles when a small CAD change, such as a bracket shift or a mounting-hole update, silently violates a requirement or interface decision made months earlier. The context that would catch it is scattered across CAD files, requirements docs, ICDs, Slack threads, and senior-engineer memory, and no current tool keeps a live link between a CAD feature and the constraint it satisfies. Pre-freeze decisions drive 86% of program rework cost and late changes cost 13x more to fix.

Satellite photo trails are turning from a nuisance into systematic error for professional surveys, space telescopes, and amateur astrophotography. The Low Earth Orbit population is around 13,000 satellites today and is projected to reach 58,000 to 100,000 or more by 2030, and contamination is already measured: at the Zwicky Transient Facility, twilight streaks rose from under 0.5% to about 18% in 21 months. The harm is unrecoverable, since missed first detections of transients, near-Earth objects, and supernovae cannot be recaptured, yet no operational fleet meets the IAU magnitude-7 brightness limit and today’s workarounds (frame rejection, streak masking, hardware dimming, schedule-dodging) only treat symptoms and do not scale.

The project focuses on the challenge of illegal fishing in the Eastern Tropical Pacific Marine Corridor (CMAR), particularly around the Galápagos Islands, where industrial vessels can deliberately switch off their tracking systems and operate undetected inside marine protected areas. Although authorities currently rely on AIS tracking, satellite monitoring, and physical patrols, these approaches have significant limitations when it comes to detecting so-called “dark vessels” and smaller boats that fall outside conventional surveillance systems. The study identifies a key monitoring gap: the dependence on data voluntarily transmitted by vessels themselves and the inability to continuously monitor vast ocean territories. As a result, critical migratory species such as sharks, sea turtles, rays, and marine mammals remain exposed to illegal fishing pressure, generating major ecological and economic impacts on biodiversity, tourism, and coastal communities.

The analysis concludes that the most promising intervention lies in combining multiple independent observation sources—particularly AIS data, SAR satellite imagery, night-light detection, and AI-based behavioral analysis—to automatically identify non-cooperative vessels and suspicious fishing activities in near real time. In the Galápagos case, the study also highlights the “Blue Paradox,” whereby the announcement of new marine protected areas can trigger temporary increases in fishing activity before restrictions take effect. As a potential solution, the project proposes an integrated monitoring system capable of detecting both dark fleets and small artisanal vessels, supported by automated alerts and faster operational responses from enforcement authorities. The ultimate goal is to strengthen marine conservation, protect migratory species of high ecological and economic value, and safeguard sustainable economic activities such as wildlife tourism, whose long-term value greatly exceeds that of extractive fishing.