Machine Learning Data Scientist

RITVIK RAMESH PALVANKAR

DATA SCIENTIST

Cincinnati, OH +1-404-***-**** ****************@*****.*** LinkedIn GitHub PROFESSIONAL SUMMARY

Data Scientist with 4+ years of experience in machine learning, computer vision, and large-scale data analysis. Proven expertise in building high-accuracy models (95% prediction accuracy with XGBoost), optimizing ETL pipelines (15% faster processing with PySpark), and deploying cloud-based solutions (AWS/Azure). Skilled in Python, SQL, TensorFlow, and PySpark, with a strong track record of delivering actionable insights to cross-functional teams. Passionate about solving complex problems through data-driven innovation. EDUCATION

Masters of Science in Electrical and Computer Engineering University of Florida - Gainesville, FL Aug 2019 – May 2021 Coursework: Machine Learning, Deep Learning, Digital Signal Processing, Neural Networks. SKILLS

Programming languages and frameworks :Python, MATLAB, Java, SQL, Django, Angular, Spring Boot, JavaScript Data Analysis and Machine Learning: Scikit-learn, XGBoost, TensorFlow, Supervised, Unsupervised, Spark, NumPy, Scikit learn, Pandas, Regression, Neural Networks, Time Series Forecasting GenAI: LLMs, GPT, RAG, LangChain, Hugging Face, Diffusion Models Computer Vision: Image Processing, OpenCV, TensorFlow, Deep Learning Database: MySQL, PostgreSQL, DB2

Cloud Computing: Amazon Web Services (AWS), AWS S3, AWS Sagemaker, EC2, Azure DevOps, Azure Synapse Studio Development & visualization tools: IDEs like Visual Studio, Eclipse, Jupyter Notebook, Power BI Version control: Git, GitHub

EXPERIENCE

Developer, Department of Public Services, TCS, OH - USA Jul 2023 – Present

• Developing user-friendly interfaces using Angular and collaborating with cross-functional teams to integrate backend and front-end components ensuring a unified and responsive application.

• Leveraging Spring Boot for backend development achieving 10% improvement in server response times and overall performance.

• Crafting highly efficient SQL queries in DBeaver significantly reducing query response times.

• Proficiently utilizing Git for version control for team management and resolving merge conflicts with a 98% success rate.

• Resolving and managing over 25 critical bugs using Azure DevOps and documenting fixes for knowledge sharing and future reference.

• Resolved 200+ bugs and feature requests in Azure DevOps with a 98% on-time resolution rate, enhancing system stability during UAT and production rollout.

• Interacting with business and stakeholders to understand requirements, addressing concerns and ensuring timely delivery of fixes and feature updates.

Data Scientist, FedEx, TCS, TN - USA Dec 2021 – May 2023

• Processed over 40 million rows using PySpark, reducing data analysis time by 15%.

• Build and optimized data pipelines using Azure Synapse studio, reducing the query latency by 18 %.

• Evaluated models using performance metrics (AUC-ROC, F1-score, precision-recall curves) and deployed solutions to production environments.

• Developed and fine-tuned predictive models using Python and PySpark to classify defaulters and non-defaulters, achieving an 87% model accuracy in credit risk segmentation.

• Assisted in deploying machine learning models into production and validated results against business KPIs, improving early defaulter detection by 20%.

• Developed comprehensive documentation of analysis, data preparation, model development, and validation processes, facilitating knowledge transfer and replication of results.

• Delivered actionable insights through client presentations, increasing satisfaction by 10% satisfaction. Data Science intern, Shoptaki, NY -USA Jun 2021 - Nov 2021

• Developed a CNN-based image analysis system for blockchain transactions, improving accuracy by 6% using TensorFlow and data augmentation techniques.

• Implemented anomaly detection algorithms (Z-score, IQR) to identify 1.2% high-risk transactions, boosting fraud detection recall by 25%.

• Applied anomaly detection (Isolation Forest, Autoencoders) to flag 1.2% fraudulent transactions. Later explored LLM-powered transaction categorization to reduce manual labelling effort by 30%.

• Built real-time dashboards (Power BI) for fraud monitoring, reducing manual review time by 40% through automated alerts and visual analytics.

Contact this candidate