Pramod Ch

Technical expertise and spearheaded POC projects for data engineering, Analytics and AI, establishing myself as the department's AI implementation expert. Streamlined historical big data processing with Databricks & Delta Live Tables. Collaborated with stakeholders to translate complex data insights into compelling data visualizations and increased user engagement by 6%. Developed an anomaly detection pipeline for vehicle data for last 2 years, enabling proactive identification of potential issues. Leveraged Explainable AI (XAI) to identify key features contribute to real-world fuel inefficiency, enabling targeted optimization strategies. Built an inhouse object detection model for vehicle telltale detection as a part of Adrenox Mlens android and iOS app leading to a 75% reduction in development time and 90% reduction in cost. Developed a time series forecasting model to estimate the demand of spare parts requirement to help the supply chain and logistics. Leading development of a persona-based RAG LLM chatbot for Mahindra, empowering product, sales, and design teams with customer-centric insights from vehicle reviews. Tools: Databricks, Pyspark, SQL, GCP Vertex AI, PowerBI, Qliksense, mlflow, gradio, Langchain, Huggingface.

Contributed to enhancing the testing efficacy of the tractor display system by effectively identifying areas for improvement and proposing data-driven solutions.

Leveraged data analytics on user experience data of vehicle's infotainment system, particularly in climate controls and navigation, achieving a 12% improvement. Enhanced ETL processes through optimized Glue job scripts, ensuring smooth data flow for analysis. Analyzed infotainment usage data to address complex business use cases. Identified potential KPI (re-)calculations and creation of insightful dashboards and reports. Tools: PySpark, Python 3, MySQL, AWS (S3, Sagemaker, Glue, Redshift), Tableau.

Developed robust and real-time visualization systems that depict gaze vectors from both head and eyes, leading to a 15% improvement in identifying driver focus areas. Tools: C++, Python, OpenGL, Atlassian tools (Jira, Confluence).

Designed a data pipeline and ML algorithm for monitoring vital signs (heart rate and breath rate) using radar for vehicle interior monitoring. Performed data pre-processing, model selection, and evaluation, validating results against existing solutions to ensure reliability. Tools: Python 3, Tensorflow 2, Matlab, Simulink, MLflow, Linux.

Designed and implemented data pipelines and ML algorithms for large-scale vehicle CAN bus data, enabling accurate drivable path estimation models. Developed a novel unsupervised clustering method, followed by training a neural network, which increased model training efficiency by 29%. Conducted a comprehensive literature review to evaluate and integrate novelty detection techniques for data pipeline. Tools: Python 3, TensorFlow, Keras, Numpy, SciKit-learn, Pandas, bokeh, Matplotlib, Seaborn, Plotly (Dashboard), Atlassian tools, Airflow and ADTF.

Analyzed production data in order to increase man, machine and material productivity.

Focus: Machine learning, Data Science, ADAS & Autonomous driving. Thesis: Multi-object tracking and interactive event prediction in traffic scenes using Graph Neural Networks.