At Unit 1991, I designed and implemented a constraint-satisfaction engine using genetic algorithms and heuristics to automate complex, multi-variable event scheduling.
I also built an end-to-end pipeline for clustering unstructured textual datasets. It used BERT embeddings for semantic similarity and tree-based algorithms for classification, achieving over 80% accuracy.
I engineered a post-processing ML pipeline to parse, clean, and reconstruct spatial and semantic structures from noisy OCR text outputs.
I'm committed to applying machine learning and deep learning methodologies to complex problems and measurable business outcomes.
