Aditya Rajeev

I'm a fourth-year Computer Science (Specialist) undergraduate at the University of Toronto, focused on the intersection of computer systems and machine learning. I believe that the future of AI lies in understanding the low-level technology that powers our AI systems.

Currently, I'm a Research Assistant at the embARC Lab with Prof. Nandita Vijaykumar, funded by the Department of Computer Science, developing optimization frameworks for GPU kernels for cryptographic workloads.

Previously, I worked as a Software Engineering Intern at Mezzi and Tigera, where I built scalable backend systems and AI-powered tooling. My research has spanned GPU security, neuro-inspired memory models, and medical imaging ML.

I hold a 3.99/4.0 GPA (Dean's List).

Academia

2026–present

embARC Lab

With Prof. Nandita Vijaykumar. Building a zkML Transformer Hardware Optimization Framework — writing GPU kernels for finite field arithmetic and hardware-software codesign infrastructure for cryptographic workloads.

2026

SITH Lab

With Prof. Gururaj Shaileshwar. GPU/security research on Rowhammer on AMD GPUs. Under review at ACM CCS 2026.

2025–26

Computational Neuroscience Lab

With Prof. Brokoslaw Laschowski. Building neuro-inspired memory models. Findings under review at NeurIPS; reviewer at NeurIPS 2026.

Winter 2026

Teaching Assistant, Systems Programming (CSC209)

2024–25

Vector Institute / UHN

With Prof. Michael Brudno. Building ML models for pneumothorax detection.

Awards:

• Department of Computer Science Award ($12,000, 2026)
• University of Toronto Excellence Award ($7,500, 2024)
• Dean's List

Industry

Coursework

• Neural Networks & Deep Learning CSC413
• Introduction to Machine Learning CSC311
• Operating Systems Design CSC469
• Operating Systems CSC369
• Compilers CSC488
• Computer Architecture CSC368
• Algorithm Design CSC373
• Data Structures & Algorithms CSC263
• Introduction to Databases CSC343
• Computer Organization CSC258
• Systems Programming CSC209
• Linear Algebra MAT223

Projects

• Train in a Snap - AR emergency training with Snap Spectacles, integrating Snap3D and Vapi Voice AI over WebSockets (Hack the North semifinalist, top 12% of 256 teams)
• Mini-Dynamo - Fault-tolerant key-value store in C with consistent hashing, vector clocks, and a POSIX multi-threaded networking engine
• mAIstro - Full-stack RAG engine with cross-document indexing across PDF/DOCX and a spaced-repetition scheduler (GenAI Genesis)