Deepak Yadav

Experience

• 

  IIEST, Shibpur

  Jul 2018 - Jun 2019

  Postgraduate Researcher

  Advanced Research in Wireless Sensor Networks (WSNs):* Conducted an in-depth exploration of fault-detection methodologies in WSNs, delving into statistical-based and machine-learning-based approaches.* Engineered a sophisticated algorithm dedicated to detecting faulty nodes within Wireless Sensor Networks (WSNs), contributing to the advancement of fault-tolerant systems.

• 

  Activa Inc

  Aug 2019 - Apr 2021

  Project Overview:* Led the creation of an innovative AutoML platform, enabling users to effortlessly train machine learning models by uploading diverse datasets (images, text, time series). The user-friendly interface simplifies the process, automating model selection and training at the click of a button.Key Contributions:* Designed and implemented a comprehensive AutoML pipeline covering data preprocessing, feature engineering, model selection, and hyperparameter tuning.* Introduced a genetic algorithm-based approach for hyperparameter selection, enhancing model training efficiency by automating optimal hyperparameter search.* Ensured platform scalability and robustness, accommodating varied datasets and delivering high-quality machine-learning models.Achievements:* Successfully developed an AutoML platform, democratizing AI by simplifying machine learning model training.* Drastically reduced time and effort for model development through automated hyperparameter selection, enhancing efficiency.* Facilitated real-world application deployment by providing downloadable models and cloud inference endpoints. Show less Project Description:* Successfully completed a 5-month internship at Activa, where the primary focus was to develop a PoC for United Technologies Corp. (UTC) to address the challenge of automating the detection of internal cracks in aircraft engines.Achievements:* Developed a Mask-RCNN-based computer vision model for precise internal crack detection in aircraft engines.* Applied deep learning techniques, significantly boosting accuracy and efficiency in crack detection.* Collaborated seamlessly with a diverse team of engineers to ensure integration with UTC's specified requirements.* Successfully pitched the Proof of Concept at T-Hub Hyderabad, showcasing its potential to automate and enhance efficiency in aircraft maintenance processes. Show less

  • Associate Data Scientist

    Dec 2019 - Apr 2021

  • Data Science Intern

    Aug 2019 - Dec 2019
• 

  Sony India Software Centre

  Jun 2021 - now

  Consultant (AI/ML)

  * Focusing on the implementation of edge AI using Sony's IMX500 device.* Evaluated TensorFlow, Keras, and PyTorch layers for compatibility and performance on the IMX500 hardware, ensuring seamless integration of machine learning models.* Conducted an exhaustive analysis of model performance across various types, including float, quantized, and edge formats, optimizing inference processes.* Developed a tailored dashboard with Streamlit and a Python backend, streamlining the evaluation of model layers. This tool enabled rapid performance comparisons, simplifying decision-making. Show less