Work on Sina's Page

Cell2Sentence

Mon, 01 Jan 0001 00:00:00 +0000

Cell2Sentence explores how biological measurements can be represented as sequences that language models can process.

The project adapts LLM ideas to single-cell biology, connecting representation learning, natural language modeling, and biological discovery.

See the paper and code for technical details.

Machine Learning for Protein Biology

Mon, 01 Jan 0001 00:00:00 +0000

During my PhD, I worked on machine-learning methods that infer biological and structural properties directly from protein sequence.

The work covered several related problems: predicting intrinsically disordered regions and their functions, DNA-binding residues, protein-binding residues, crystal-structure quality, and sequence-derived signals associated with druggability.

A recurring theme was learning useful representations from sequence when direct structural evidence is sparse or unavailable. That meant thinking carefully about labels, evaluation, biological priors, and the gap between benchmark performance and useful scientific inference.

Patient Record Linkage at Scale

Mon, 01 Jan 0001 00:00:00 +0000

Overview #

I designed and built a large-scale patient matching system combining learned representations with deterministic constraints.

The system used neural embeddings, blocking strategies, clustering, and precision/recall evaluation techniques to link records across healthcare datasets while balancing accuracy and scalability.

Why it matters #

Entity resolution is a foundation problem for healthcare data. Small errors have large downstream effects, and the system must combine statistical learning with domain knowledge.