Pranay Raj Sarap

Experience

• 

  Mindtrail Technologies PVT LTD

  Feb 2019 - Aug 2021

  Data Associate

  1. Built churn prediction and sales forecasting models with machine learning algorithms, achieving 25% higher accuracy and reducing false positives by 15%.2. Ensured data quality and integrity through advanced preprocessing techniques, including imputation and outlier detection.3. Developed interactive Tableau dashboards to visualize sales trends and customer behaviors, improving stakeholder decision-making efficiency.

• 

  Cognizant

  Sept 2021 - Jun 2022

  Programmer Analyst

  1. Improved customer segmentation accuracy by 20% through the implementation of PCA and Gaussian Mixture Models.2. Conducted advanced analytics, including correlation and hypothesis testing, to identify spending patterns and optimize pricing strategies, increasing revenue by 15%.3. Designed a high-value customer loyalty program by reducing data dimensionality using PCA and uncovering critical customer features.

• 

  Wichita State University

  Feb 2023 - Sept 2023

  Data Associate

  1. Deployed LangChain and AWS SageMaker to build NLP models for summarization and sentiment analysis, reducing training time by 20%.2. Streamlined deployment workflows using AWS Bedrock and API Gateway, improving access to pre-trained models for various NLP tasks.3. Enhanced chatbot coherence by leveraging Jurassic-2, Titan Text, and Claude LLMs, achieving a 15% improvement in human evaluations.

• 

  Capital One

  Sept 2023 - Dec 2023

  Jr. Data Associate

  1. Optimized chatbot interactions by leveraging LangChain’s framework for multi-LLM orchestration, improving scalability by 15%.2. Designed and implemented multi-step LLM workflows, including sentiment analysis and database queries, achieving 20% higher response accuracy.3. Created customized prompt templates to handle ambiguous inputs, boosting user engagement by 25%.

• 

  Lisnen

  Dec 2023 - May 2024

  NLP Engineer

  1. Gained expertise in deep learning frameworks (CNNs, RNNs, attention mechanisms) through the analysis of 50+ audio classification research papers.2. Enhanced sound classification accuracy by 30% using data augmentation techniques and engineered low-latency audio feature extraction pipelines.3. Designed real-time anomaly detection systems for critical use cases like detecting baby cries, doorbells, and fire alarms.

• 

  Life Enterprises, Inc.

  Sept 2024 - now

  Data Scientist

  1. Analyzed user-uploaded data using Python, Pandas, and scikit-learn to uncover trends and insights, driving data-informed decision-making.2. Built robust data pipelines leveraging Google Cloud tools (BigQuery, Firestore) to streamline data retrieval and processing.3. Developed a personalized recommendation system using collaborative and content-based filtering techniques for enhanced user experience.