At BrainChip, I worked on building machine models practical for edge deployment by fine-tuning MobileNetV2 and MobileNetV4, optimizing them with quantization, pruning, and staged transfer learning.

I built reproducible pipelines, analyzed power, memory, and speed trade-offs, and re-engineered MobileNetV4 for Akida-compatible conversion, ensuring efficient and reliable on-device performance.

• Co-Director of the AI & Innovation Team at GRIDS Club
• Coursework: Foundations of AI, Machine Learning, Information Retrieval, Deep Learning, Natural Language Processing, Analysis of Algorithms, Database Systems

✧ Proud Trojan, building ideas as much as models ✧

Developed an AI interview assistant that infers Big Five personality traits from facial and text cues, helping candidates and interviewers gain deeper insights. Built with TCNs, OpenFace, and MediaPipe, and deployed as FastAPI services on AWS for scalable, real-time assessments.

Explored how clustering problems can be reframed as SAT formulations, reducing Minimum Sum Diameter Clustering into 3-cluster constraint cases. Applied the DPLL algorithm across 2-SAT and 3-SAT, while connecting ideas from Knuth’s SAT theory and Psat models to concepts like sparsity in ML.

Solid CS base; discovered a passion for ML.