Angelic Charles

Angelic Charles

@visionbyangelic

Computational Neuroscience ยท ML
Probing how intelligence emerges in brains and machines from the ground up. ๐Ÿง 

B.Sc. Biochemistry B.Sc. Software Engineering Computational Neuroscience
Synaptic Soundscape Hover background to play
๐Ÿ’ฌ Leave a Note ๐Ÿ—“๏ธ Book a Call
Hint: Double-click avatar for brain activity.
๐Ÿชฐ
โ†’
Closed-loop simulation coupling complete Drosophila connectomes (FlyWire, BANC) with Brian2 point-neuron dynamics and an embodied biomechanical body (MuJoCo, flygym) to evaluate emergent locomotion, grooming, and feeding without hand-coded motor rules.
Research Question: Can natural coordinated behaviors emerge organically when real synaptic wiring is given simple biophysical rules and placed in an embodied physical simulation?
Research Stack: Brian2 ยท FlyWire / BANC ยท MuJoCo ยท flygym ยท Python
๐Ÿ”ฌ
Completed โ†’
Computational study examining how biological constraints (spiking thresholds, metabolic sparsity penalties, and recurrence) shape internal spatial representations, place-cell-like geometry, and zero-shot transfer stability in embodied agents under severe sensory ambiguity.
Key Finding: Constrained spiking networks produced place-cell-like coding with up to ~80x higher spatial information per spike than unconstrained baselines and demonstrated superior transferability into continuous 3D environments.
Research Stack: Python 3.10+ ยท snnTorch ยท PyTorch ยท Blender 3D ยท MiniGrid ยท PPO
Neuro & Simulation
Brian2, MuJoCo, flygym, snnTorch, MNE-Python, FreeSurfer
Machine Learning
PyTorch, TensorFlow, Scikit-Learn, PPO, Spiking Networks
Languages
Python, SQL, JavaScript, TypeScript, C++, Java
Systems & Tools
Blender 3D, FastAPI, Streamlit, Docker, Git, Arduino