MS

Mayur Shimbre

Data Scientist II · ML Forecasting & GenAI

Bengaluru, Karnataka, India

@mayurshimbre25474460

0 followers

About

Data Scientist II at Honeywell building ML forecasting pipelines on Azure Databricks and PySpark, with experience improving forecast accuracy and reducing runtime. Also develops Python-based LLM and RAG applications using IBM watsonx.ai, Gradio and LangChain.

Experience

  • Data Scientist II
    HONEYWELL
    Jul 2024 – Present

    Improved global-services revenue forecast accuracy from 80% to 95% using business-driver features in SARIMAX across an 18-month horizon. Built Azure Databricks/PySpark pipelines for transaction-level profiling, validation, cleansing, feature engineering, modeling and backtesting. Packaged reusable Python wheels with entry points and dynamic forecast parameters; used GitHub Actions for Databricks DEV builds. Orchestrated preprocessing, training, evaluation and reporting in Databricks Jobs; cut end-to-end runtime from 8 to 1 hour (87%). Connected explainable forecasts to business KPIs and supply-chain drivers, partnering with cross-functional teams on planning.

  • Deep Learning Instructor
    IIT Madras
    Jan 2024 – Jul 2024

    Designed and delivered deep learning lectures on CNNs, RNNs, LSTMs and Transformers, supported by PyTorch demonstrations. Built ML models to predict student performance and identify factors influencing academic outcomes for targeted support. Mentored applied deep-learning projects spanning data preparation, model experiments, evaluation and implementation.

Education

  • B.Tech + M.Tech
    Biological Engineering (Computational Biology) · 2023

Skills

Tools / apps / platforms

2 to 5 years 2 to 5 years Several years of experience.
GitHub GitHub Actions
<3 months <3 months Just getting started - under three months. This is also what shows when a level has not been set.
Git Docker

Projects

  • LLM & RAG Applications
    IBM watsonx.ai, Gradio, Python, LangChain

    Built document Q&A, financial scenario analysis and AI interview practice interfaces with retrieval and source-evidence views.

  • Tweet Sentiment Analysis
    NLTK, Logistic Regression, Python

    NLP classification project using tweet preprocessing, frequency-based sentiment features and logistic regression to classify polarity.

  • Synthetic Text Generation
    GRU, Python

    Character-level language model for Shakespeare text generation using embeddings and recurrent layers, with sampling and checkpoint evaluation.

Explore the Broxer community

Online · instant AI help
Broxer