Arun Kumar Reddy Kandula

About Arun Kumar Reddy Kandula

Experienced in Computer Vision, Deep Learning and Machine Learning. I am looking forward…

Experience

1. Data Scientist i

   IFF

   Jun 2024 — Present · DE, US

   Project 1: Auto GC. This project aims to build a robust algorithm to detect the chemical compounds in the Gas Chromatography signals (Time-series data). • Adapted and enhanced the WaveGAN architecture to generate synthetic Gas Chromatography peak signals, eliminating manual labeling. Trained the model on SageMaker using the PyTorch framework.• Implementing a classifier to detect and identify peak classes using generated synthetic data, currently supporting 10 distinct peak types. • Addressing the limitations of existing ML-based and rule-based methods by leveraging a deep learning approach that utilizes all features, enhancing robustness and accuracy, especially for edge cases. Project 2: SOP Distiller. This project aims to distill long Standard Operating Procedures (SOPs) into checklists to help new employees in manufacturing sites quickly learn workflows and machine operations. • Generating checklists from existing SOP documents using Large Language Models (LLMs) like GPT-4o, simplifying long paragraphs into easy-to-follow steps using PromptFlow on Azure. • Extractracted and parsed text from various document formats (.docx.doc.docm, pdf) using custom-built methods, structuring the data for effective LLM processing.• Developed an automated agentic evaluation framework and self-improvement loop using LangChain, enhancing the accuracy of generated checklists.

Education

• Asian Institute of Technology

  Master's degree, Microelectronics and Embedded Systems

  2017 — 2019

• Jawaharlal Nehru Technological University

  Bachelor of Technology - BTech, Electronics and Communication Engineering

• University of North Texas

  Master's degree, Artificial Intelligence