High school student · Self-directed researcher · AI × Spacecraft Control
I am fascinated by how a spacecraft survives in silence —
and how intelligence, embedded in software, might become its second heartbeat.
I explore this question by building reproducible simulations, controllers, and logs,
treating learning itself as an engineering process rather than a competition.
Trajectories drift. Controllers adapt.
Failure is telemetry.
A long-term, self-directed research project focused on
AI-controlled orbital propulsion systems in simulation.
This project is not a demo.
It is an evolving system designed with engineering rigor and traceability.
- Physics-based orbital dynamics and propulsion modeling
- Expert (rule-based) controllers as stable baselines
- Transition to imitation learning and reinforcement learning
- Decision-making over:
- thrust ignition
- thrust magnitude
- impulse vs continuous propulsion
- Long-horizon rollouts under:
- noisy inputs
- partial actuator faults
- imperfect state information
- Every run produces logs treated as mission data
- Results are reproducible and auditable
- Failures are recorded, categorized, and analyzed
- Progress is measured by stability and insight, not only reward curves
This work produced my first reproducible orbital transfer simulation
and a stabilized expert baseline, validated across multiple orbital conditions.
🔗 https://github.com/Sean-ZhiXin-Li/spacecraft-ai-controller
Every autonomous system stands on layers of engineering.
Alongside my main project, I build foundational skills across disciplines that support spacecraft autonomy:
- 🤖 Robotics and control experiments
- ⚡ Embedded systems (Arduino, sensors, low-level interfaces)
- 🛰️ CubeSat structural modeling and CAD exploration
- 🎮 Reinforcement learning from simple environments to orbital-scale tasks
🔗 https://github.com/Sean-ZhiXin-Li/tech-foundations
A process-focused engineering repository designed to build
long-term, lab-ready capability, not showcase results.
This repository is not a research project and not a product.
It exists to answer one question clearly:
Can I independently build, debug, and iterate on a Python + AI engineering system
in a Linux environment, from a blank file?
- Python engineering from scratch (no templates)
- Linux / WSL–based workflow
- Git discipline and reproducibility
- Debugging as a first-class skill
- Clear execution → metrics → verification pipeline
Each run leaves persistent, inspectable traces.
Failures are preserved, categorized, and verified — never hidden.
This repository supports future research work by strengthening
engineering habits that scale to collaborative lab environments.
🔗 https://github.com/Sean-ZhiXin-Li/engineering_ai_playground
For me, research is not only about results.
It is about process integrity.
- Commits are signals: proof that the system is still alive
- Bugs are disturbances that reveal structure
- Logs are navigation charts through uncertainty
- Progress is often spiral, not linear
Traditional drills and competitions exhausted me.
Open-ended projects taught me persistence.
Even when today’s trajectory is unclear,
the learning system continues to integrate.
I am curious about the universe —
about regions no probe has reached and signals that have gone silent.
Each lost spacecraft feels like a failure of endurance.
That feeling pushes me to design controllers that adapt longer,
degrade gracefully, and survive uncertainty.
I do not attempt to master all of physics or AI.
I focus only on what matters for spacecraft autonomy,
assembling knowledge module by module.
This is not a promise.
It is a direction vector.
- 🌌 Age 16 — Wrote my first orbital simulation out of curiosity
- 🌱 Now — Building reproducible AI-controlled systems and engineering pipelines on GitHub
- 🛠️ Early 20s — Deepen foundations in engineering, AI, and control
- 🚀 Long-term — Contribute controllers tested beyond simulation
The exact path is unknown.
The mission continues anyway.
- GitHub: https://github.com/Sean-ZhiXin-Li
- Email: tlizxin209625@gmail.com
Total Contributions: 205 | Longest Active Period: 45 days | Activity Span: Sep 2024 → Present
https://github-readme-stats.vercel.app/api/top-langs/?username=Sean-ZhiXin-Li&layout=compact