I’m a computer science student focused on deep learning research. I work on language and multimodal models,
with a particular interest in how they learn, adapt, and behave.
Architected and trained TrooperViz, a multimodal safety guardrail model built with Python, PyTorch, and Hugging Face Transformers, enabling evaluation of text, image-text inputs, multi-turn conversations, and assistant-generated outputs.
Fine-tuned a vision-language model for multimodal safety classification and merged it with a text-only guardrail model using SLERP, creating one system for visual, conversational, and generative-AI interactions.
Built data and evaluation workflows for annotation, preprocessing, synthetic conversation generation, benchmarking, and interactive inference with Gradio and vLLM.
Architected an LLM-powered NL-to-SQL RAG pipeline with few-shot prompting, input normalization, parameterized SQL sanitization, session memory, and W&B observability across five Azure ML endpoints.
Built an SVD collaborative-filtering recommendation pipeline and an LLM-driven taste profiler for nightly user preference generation.
Engineered recipe ingestion using Scrapy, Instagram and TikTok integrations, GPT-4o-mini classification, language filtering, and audio transcription fallback.
