Bangladeshi Students Build Satellite-Risk Tool and Take a SpaceXAI Prize at HackMIT

Two students from Bangladesh have taken one of the sponsor prizes at HackMIT 2026 after building a prototype that turns raw orbital data into a readable picture of collision risk.

Ashraful Islam Tutul and Md. Tanvir Ibn Alam, both enrolled at Truman State University in the United States, won the SpaceXAI "Make It Legendary" challenge with a project they call Scutaris. Their entry was one of two selected in that category. HackMIT, held at the Massachusetts Institute of Technology, drew more than 1,000 participants this year and ran 23 sponsor tracks.

The pair describe Scutaris as a command view for orbital safety. It pulls in more than 9,000 live satellite and debris records from CelesTrak, stores them in Elasticsearch, and lets a user ask questions in ordinary language. The system then ranks likely threats. Once a user picks a case, Grok drafts a short briefing on what the data shows and what the limits of that reading are. Grok Imagine can turn the same case into a shareable risk card. A 3D Earth shows satellites, debris, and their paths, along with how close objects come, the estimated chance of a collision, and when the nearest pass is expected.

The split of work followed their degrees. Ashraful, a computer science student with a data science minor, built the interface, the 3D view, live search, the Grok Imagine link, and the demo. Tanvir, studying data science with an astronomy minor, handled the backend, orbital processing, the search index, risk screening, and a five-stage Grok pipeline. They designed the idea and the final integration together. Tanvir's astronomy training shaped which fields they kept; Ashraful focused on making the crowded sky understandable without forcing a user through thousands of raw rows.

They started by reading the SpaceXAI and Elastic briefs and decided the sponsor tools had to sit at the centre of the product, not on the side. Public orbital data came first, then search and screening, then the API, the globe, and the AI layer. Ashraful said they finished by wiring Grok and Grok Imagine into the full flow and fixing breaks before the deadline.

The hard parts were keeping thousands of records consistent and searchable, keeping the 3D view responsive, and joining live search, threat selection, analysis, and image generation without the chain falling apart. They split remaining tasks, tested pieces alone, then assembled the system. In front of judges they also showed how Grok, Grok Imagine, and Cursor were used while the tool was being built.

Scutaris is still a hackathon prototype, not a system that could guide a satellite. The students say any operational version would need authoritative data, heavy testing, and people who work in orbital mechanics and satellite operations. They want tighter path and closest-approach maths, better risk estimates, official safety feeds, real-time alerts, historical review, shared incident workspaces, and a human sign-off before any manoeuvre is recommended.

The result is a small but clear signal: Bangladeshi students competing at MIT can turn public space data and current AI tools into something judges in a SpaceX-linked track chose to reward.