Abhinav Bohra is a Senior Applied Scientist with over 10 years of experience building production-scale ML/AI systems. At Amazon, he leads the development of self-supervised attribute extraction frameworks processing more than a billion products using teacher-student distillation. His expertise is in LLM fine-tuning and distillation (LoRA/QLoRA), Two-Tower neural network architectures for recommendation, multimodal AI systems combining LLMs with computer vision and low-latency LLM inference. His recent work includes explainable LLM-based recommendation engines addressing cold-start challenges, multimodal pricing anomaly detection using LLMs and DinoV2, and Quantile Regression Neural Networks for uncertainty quantification. His most recent work has been published at AACL 2025 on Quantum NLP.

His expertise is in following areas:

1. Recommendation and Ranking models: Two-Tower models, Multimodal ML systems (LLMs + vision embeddings + text embeddings + gated fusion)
2. LLMs: Fine tuning, Teacher-Student distillation, LLM explainability, Parallel decoding and low-latency inference
3. Quantile Regression Neural Networks