Ashish Lathkar

🚛 Building Trucking Future!

🧭 Building Roadmaps!

✦ Developing a feature for NcDashboard that enables users to interact with NetCDF datasets via natural language queries, using Python, xarray, and numpy for processing and analysis. ✦ Incorporating a Large Language Model (LLM) through LangChain to generate Python code dynamically based on user queries, facilitating complex data computations and analyses. ✦ Enhancing NcDashboard’s functionality by automating code generation, making advanced data analysis more accessible to users with diverse technical backgrounds.

✦ Enhanced a text-to-text (T2T) & voice-to-voice (V2V) AI agents, boosting efficiency by 22% through granular transcript analysis & cutting inference latency by 33%; improved accuracy by 12% using GPT-4, AWS Lambda, Twilio Voice API, & FastAPI while debugging Python scripts for system reliability. ✦ Worked on GraphRAG-powered chatbot that constructs a dynamic scalable knowledge graph from client FAQs, enabling real-time query resolution & automated scheduling; accelerated response times by 40% with GPT-4, LangChain, AWS, Milvus, Neo4j, and FastAPI, ensuring seamless data retrieval. ✦ Engineered a containerized chatbot system with React, Node.js, Docker, & Kubernetes, optimizing scalability, high availability, & real-time interactions, ensuring low- latency deployments on AWS ECS & Lambda with secure API integrations.

✦ Boosted task completion rates by 15%, increasing average tasks completed per user from 20 to 23 daily, by engineering a task management app with enhanced UI/UX and seamless cloud integration. ✦ Enhanced security with AWS Cognito, achieving a 100% secure sign-in rate and reducing unauthorized access attempts by 32%. ✦ Saved employees 1 to 1.5 hours per day by deploying the app on AWS Amplify, providing 200+ active users with high availability, remote access, and streamlined task management.

✦ Spearheaded the application of gradient boosting and SVM for advanced credit risk modeling on the Bondora platform. Conducted extensive EDA and data preprocessing using Python, handling a colossal 1 TB of data, and showcasing technical prowess. ✦ Led a team to optimize Spotify skip prediction, achieving a record 87.65% accuracy with a decision tree classifier. Conducted thorough code reviews and successfully deployed the model using the render platform. This aligns seamlessly with industry goals, showcasing a proven ability to enhance user experience through advanced predictive modeling. ✦ Pioneered cutting-edge Exploratory Data Analysis (EDA), precise data preprocessing, and meticulous data cleaning. Orchestrated the project using Jupyter Notebook, Python, Numpy, Pandas, Matplotlib, Sklearn, and an innovative suite of technologies.

✦ Developed exceptional tutorials for JEE Main, JEE Advanced, and MHT-CET competitive exams, specializing in complex topics including Linear Algebra, Calculus, Probability and statistics, Gradient Descent, and Graph theory. ✦ Achieved a remarkable 10% enhancement in results, particularly benefiting students facing behavioral challenges. Focused on fostering success through engaging teaching methodologies. Team collaboration, internet scale problems, data products ✦ Demonstrated flexibility, creative problem-solving, and effective communication—key attributes crucial for contributing to an innovative and collaborative environment. The commitment to improvement aligns with a culture of constant learning and excellence.

✦ Engineered a groundbreaking H1-B visa approval prediction model, leveraging advanced techniques like logistic regression and naive Bayes algorithms, showcasing expertise in data science tools including Python, Jupyter Notebook, and Sklearn. ✦ Streamlined data collection, cleaning, and analysis processes, employing advanced feature engineering to pinpoint crucial model features. Resulted in a significant 5% increase in F1-Score, showcasing a tangible impact on decision-making accuracy. ✦ Translates passion for data science into actionable insights, reflecting a commitment to data-driven decisions—a crucial aspect in enhancing algorithms and optimizing services for global users.