📧 Connect with me on LinkedIn | ✉️ Email: atharva.bhusari@rutgers.edu
🎓 Master of Science in Data Science | Rutgers University
📊 Trainee Business Analyst @ R S Infocn. Inc.
- Designed, developed, and enhanced JD Edwards (JDE) applications and reports by translating business and functional requirements into scalable ERP technical solutions
- Worked closely with cross-functional business teams across Order-to-Cash (O2C), and Freight Optimization to align JD Edwards functionality with operational workflows
- Built Orchestrations using the JDE Orchestrator tool to load and extract transactional data, helping reduce data handoff time to downstream systems by ~30%
- Migrated legacy IBM Showcase reports to ReportsNow, improving report performance, reducing maintenance effort by ~35%
- Partnered with functional analysts and business users to perform unit testing, system integration testing (SIT), and user acceptance testing (UAT), resulting in smoother releases and fewer post–go-live issues
- Supported JD Edwards system integrations and data validation activities, contributing to more reliable reporting and visibility
🩺 AI/Backend Engineer @ Curanostics
- Built and refined ETL pipelines to extract and transform unstructured EPIC EHR data into structured JSON formats, supporting downstream analytics and model development with high data accuracy
- Assisted in implementing HIPAA-compliant data processing workflows, ensuring secure handling of PHI and adherence to healthcare regulatory standards
- Developed and optimized RESTful APIs using FastAPI and PostgreSQL to enable data exchange between clinical applications and internal ML services, improving API response times by several hundred milliseconds
- Participated in deploying and maintaining a production web application on Vercel with CI/CD pipelines, contributing to high application availability and reliable performance
🎲 AI/ML Development Intern @ Entertainment Technologists
- Automated Text and Data Extraction from Receipts: Implemented a system using OCR and AI models via the Replicate API to extract and structure receipt data from PDFs and images into a predefined JSON schema, capturing key details like store name, date, items, total cost, and payment method.
- Database Integration and Real-Time Updates: Integrated MongoDB to store structured data, enabling efficient retrieval and updates. Supported interactive user edits for extracted fields through a UI, ensuring data accuracy and consistency.
- Error Handling and Versatile File Support: Designed robust error-handling mechanisms for unsupported file types, missing inputs, and API token errors. Supported various file formats, including PDFs, PNGs, and JPGs, to enhance user flexibility.
- End-to-End Workflow Implementation: Developed an end-to-end solution with a Flask-powered backend and a Vite.js frontend, allowing seamless file uploads, real-time data review, updates, and search functionality, enhancing usability and application scope.
- Project Link
🌟 Data Science Research Assistant @ Rutgers Institute for Health (New Brunswick, NJ)
- Collaborated with a research team to create a Machine Learning pipeline for the classification of over 1000 bio-markers
- Achieved a 90% improvement in pipeline efficiency through parameter fine-tuning for algorithms such as KNN, SVM, and Random Forest
- Showcased the development of a Machine Learning pipeline that enhanced the predictive analysis of bio-markers by 80%
🚀 Programmer Analyst @ Cognizant Technology Solutions (Pune, Maharashtra, India)
- Mitigated business impact by 30% through rapid response to workflow failures using PowerCenter Workflow Monitor
- Collaborated on updating ETL workflows and mappings in alignment with business requirements using PowerCenter Designer
- Efficiently managed and optimized Data Integration Services and Process Chains in SAP, consistently achieving 100% on-time task completion
- Adhered to Agile methodology and maintained comprehensive documentation using ServiceNow and HP Application Lifecycle Management Tool
📜 Publication
- IntelliGenes: A novel machine learning pipeline for biomarker discovery and predictive analysis using multi-genomic profiles
- Sherlock: An Ensemble-based Deep Learning Framework for Fake News Detection
💼 Skills:
- Data Analysis | Machine Learning | Data Visualization
- Python | R | SQL | AWS | Tableau | Informatica
📜 Certifications:
- 📚 Google (Coursera) - Data Analytics Professional Certificate
- ⚙️ Udemy - Python for Data Science and Machine Learning Bootcamp