Yadnyesh Gosavi

Seasoned and dedicated AI Engineer specializing in Natural Language Processing (NLP), seeking a challenging position to contribute extensive expertise in Large Language Models (LLMs), Deep Learning (DL) And AIOPs (MLOPs) technologies. Eager to leverage over 4.5+ years of professional experience to drive innovation in the realm of Artificial Intelligence.

Generative AI, NLP, CV, RLHF, DL, ML, AIOPs (MLOPs), Customer Analysis, Recommendation system, Big Data, Retail, E- Commerce, Sentimental analysis.

LAILA Versioning 3.0

• Successfully enhanced an existing project, reducing training time and achieving a 20% cost saving in cloud services.
• Effectively addressed challenges during versioning, ensuring successful resolution.
• Implemented ZenML, Seldon, Kubeflow, and MLflow, achieving a 30-second improvement in response time.
• Improved accuracy rate by 4% using Hugging Face (LLM) models.
• Utilized MongoDB, Streamlit, GKE, GCloud, orchestrated Kubernetes cluster, Load Balancer, and Ingress Controller.
• Monitored project performance using Grafana and Prometheus.

Internal Organizational Projects:

• Computer Vision (CV) Project for HR Department: Developed a sophisticated facial and voice recognition application for secure candidate verification during interviews.
• Chatbot Project: Pioneered an advanced chatbot leveraging LLM, OpenAI, and LangChain for seamless conversational experiences.
• Text Summarization Project: Architected a text summarization system using LLM, OpenAI, LangChain, and vector spaces for efficient data analysis.

Insurance Fraud Detection

• Spearheaded the development of a POC application to automate the manual verification process for insurance claims.
• Translated complex requirements into analytical specifications.
• Implemented Flask Framework and successfully deployed the solution on AWS (EKS).

Customer analysis

• Executed a POC for comprehensive business analysis, aiming to enhance customer satisfaction, reduce churn, and potentially increase revenue by 5-10%.
• Conducted customer segmentation based on Recency, Monetary, and Frequency (RMF) values. Constructed an XGBoost Multi-Classification model to predict Customer Lifetime Value, providing valuable insights into customer behaviour.
• Deployed classification models, including Logistic Regression, Decision Tree, and Random Forest, to predict and address Customer Churn Rate ON GCloud. Implemented A/B testing strategies for optimizing product-related decisions.

Predict Quality of Wafer Sensors

• Created as proof of concept for saving manually checking of the fault sensors and save time and resources.
• Data Loader, data transformation (Datetime), Preprocessing (KNN-Imputer), Best model finder (Random Forest, XG Boost), Flask, Dockerize, Deployed on GKE.