Hemanth Kongara

M.Tech AI

Biography

I am Hemanth Kongara, and I am currently working as a Senior Data Scientist at Apna, where I focus on building and scaling large-scale search, recommendation, and ranking systems. I have independently led initiatives across content discovery, and personalization, driving a transition from entity-based retrieval to taxonomy-aware vector retrieval using two-tower models.

My work spans cross-encoders, self-attention fusion layers, and learning-to-rank models, alongside inference optimization using ONNX to achieve significant latency improvements. I have also fine-tuned large taxonomy model using LoRA and deployed ranking systems validated through both offline metrics and online experiments.

During my master's program, my research focused on medical image analysis and explainable deep learning for Optical Coherence Tomography (OCT). This work involved advanced denoising techniques and CNN explainability methods, with publications and presentations at NeurIPS Medical Imaging Workshop, MDPI (Advances in the Diagnosis of Retinal Diseases), and a submission to IEEE ISBI.

Publications

2023

Diagnostics 23

Deep-learning-based visualization and volumetric analysis of fluid regions in optical coherence tomography scans

H. R. Kasireddy, U. R. Kallam, S. K. S. Mantrala, H. Kongara, A. Shivhare, J. Saita, and others

Diagnostics, 2023

BIB URL

@article{kasireddy2023deep,
  author = {Kasireddy, H. R. and Kallam, U. R. and Mantrala, S. K. S. and Kongara, H. and Shivhare, A. and Saita, J. and others},
  journal = {Diagnostics},
  number = {16},
  pages = {2659},
  title = {Deep-learning-based visualization and volumetric analysis of fluid regions in optical coherence tomography scans},
  url = {https://scholar.google.com/citations?view_op=view_citation&hl=en&user=1g1i1B4AAAAJ&cstart=100&pagesize=100&citation_for_view=1g1i1B4AAAAJ:OP4eGU-M3BUC},
  volume = {13},
  year = {2023}
}