At Infosys Springboard, I engineered a multilingual deep-learning pipeline using Python and Flask for sequence transformations. I also used Librosa to process audio and integrated a BART sequence-to-sequence pipeline for abstractive summarization.
I designed a handwritten text recognition system combining gated convolutions, BiLSTMs, self-attention, and CTC loss. It achieved a 1.72% Character Error Rate on the Parzival manuscript.
On my RAG project, I built a PyTorch framework to vectorize, index, and query enterprise document corpora. My custom embedding layers and similarity rankers reduced target model drift and contextual hallucination indices by 38%.
I’m studying Computer Science and Engineering at Vellore Institute of Technology, Amaravati. I placed Global Rank 832 in TCS CodeVita Season 13.
