Ninder Joshi

About Ninder Joshi

As a Data Scientist at Nagarro, I apply my expertise in data analytics, machine learning, and AWS cloud to solve complex business problems in different domains. I developed and deployed predictive models using XGBoost and naive algorithms to forecast market clearing prices, achieving an 85% pop (revenue metric). I also architected and implemented a scalable simulation pipeline using AWS state machine, Lambdas, and SQS to optimize and test different scenarios and settings to maximize revenue.Previously, I was a Data Scientist at Purchasing Power, where I created the first in-house recommendation engine using transactional and clickstream data. I designed and built various models using ALS, Meta Prod2Vec, and Prod2Vec to provide personalized and relevant recommendations to customers, capturing 20% of the revenue and 25% of the user conversions. I also proposed a POC for the marketing basket analysis using data mining algorithms such as Apriori and FPGrowth. I am proficient in Python, SQL, and Tableau, and have a strong foundation in statistical analysis.I am currently pursuing a Master of Technology degree in Data Science and Engineering at Birla Institute of Technology and Science, Pilani, where I am honing my skills and knowledge in advanced topics such as deep learning, natural language processing, and computer vision. I am passionate about using data to uncover insights that drive strategic business decisions and am always eager to collaborate and share knowledge with fellow data science professionals.

Experience

1. Senior Data Scientist

   Deloitte

   May 2024 — Present · Gurugram, IN

   DC/Item Sales Forecast Bias Detection and Adjustment System- Developed an analytical framework to identify and flag inaccuracies in DC/Item-level sales forecasts by analyzing recent sales trends, year-over-year (YoY) sales data, and forecast outputs- Addressed three key biases:(1) Under/over bias compared to recent sales,(2) YoY under/over bias against last year’s sales, and (3) High variation forecasts- Applied statistical techniques like F-Test, CoV ratio to detect significant deviations in forecasting accuracy and flag these for business adjustments- Collaborated and Designed automated data pipelines with DevOps, and DE on GCP to process and flag massive DC/Item-level forecast datasets efficiently- Published insights and flagged reports to Looker dashboards, allowing stakeholders to visualize and analyze forecasting biases for manual adjustments.Serverless LLM-Based AWS GenAI Customer Interaction System- Led a team of 4 to build a serverless LLM-based voice system using AWS Bedrock, Lex, Lambda, and other services for outreaching customers for the purpose of collection of application date, along with answering user queries- Implemented an end-to-end pipeline utilizing Connect, Lex Lambda, Step Functions, Sagemaker, and DynamoDB to automate call handling, collect user data, answer user queries in real time- Performed prompt engineering in the lex to comprehend, extract information from user response and respond using AWS Polly

Education

• Birla Institute of Technology and Science, Pilani

  Master of Technology - MTech

  2023 — 2025

• Lovely Professional University

  Bachelor of Technology - BTech

  2017 — 2021