Research

MEng, Computer Science (Specialization in AI)

Distinction: magna cum laude

BSc, Computer Science and Engineering

GPA: 3.85/4 · 3-year full-tuition scholarship

Neurosymbolic methods for controlled text generation

• Combining symbolic constraints with learned, soft steering signals to guide controllable text generation in language models.

Compute Systems Architecture

• Investigated storage behavior using an SSD simulator (SimpleSSD, C/C++), as part of memory–storage co-design research. Integrated the simulator into imec's Azure-based compute experiments dashboard.
• Modeled AI workload utilization in the SSD by decoupling target token rates from device throughput to identify storage bottlenecks and saturation points. Ran latency, bandwidth, and cache experiments across multiple representative workloads to analyze performance trends.

LLM-based Course Assistant  

• Created a university course assistant using a retrieval-augmented generation pipeline, increasing safety by enabling students to only query lecture materials.
• Built using all-MiniLM-L6-v2 embeddings with FAISS vector indexing, and integrated DeepSeek-R1-14B as the LLM for context-aware response generation, incorporating lecture materials.

Computational Approaches to Problem Solving

• Weekly recitation sessions for undergraduate students and office hours for a course on algorithmic problem solving, and Python programming.

Programming Fundamentals

• Office hours per week for an intermediate OOP course, assisting students with C++ programming.