Abhinav Arora

Bachelor's

Computer Science

New Hampshire Ball Bearings Inc.

Artificial Intelligence Engineer

• Built secure, offline RAG chatbots using LLama 3.2 LLM to help each cell quickly access and query standard procedures and documents. • Optimized inference latency using model compression techniques like quantization, cutting compute costs by 60% with no accuracy loss. • Automated 3-Way Match process using fine-tuned Mistral-7B model, improving accuracy by almost 70% and reducing manual workload. • Engineered data pipelines & warehouses to support AI inferencing, cutting reporting time from hours to seconds. • Deployed Mixtral vision model for AI inferencing to extract line-item data from unstructured Purchase orders, streamlining ERP ingestion. • Built and fine-tuned CNNs, RNNs, and Transformers inTensorFlow for classification and detection using transfer learning and data augmentation