AJIT ANAND

• Appointed as Data engineer

Technologies: Python, Django, ARIMA, SQL Server, React, FastAPI, OpenAI API, LangChain, Docker

• Developed and deployed an AI-based resume parser with custom keyword extraction, improving parsing speed by 30%
• Built an intuitive web interface using React for clients to upload and parse resumes
• Integrated OpenAI APIs for semantic parsing and enhanced candidate-job matching, increasing recruiter efficiency by 25%
• Implemented salary prediction using ARIMA on historical employee data (time series forecasting)
• Created and exposed RESTful APIs via Django REST Framework and FastAPI
• Extracted, cleaned, and transformed data from SQL Server for model training and analytics
• Leveraged LangChain for intelligent context-aware operations
• Containerized components using Docker for scalable deployment

• Built models for service success prediction, handling time, and cost forecasting using Python (ARIMA, Random Forest). Applied clustering and NLP to analyze patterns. Deployed REST APIs (FastAPI/Node.js), SQL-based data pipelines. Containerized with Docker.

• Managed data conversions, data pre-processing, and data cleanup Conducted exploratory data analysis and model training Selected suitable algorithms and assessed loan default risks Analyzed creditworthiness and bias elimination through digital footprint data. Assessed accuracy and effectiveness of new and existing data sources and data analysis techniques.

• Developed loan default prediction processes Conducted customer segmentation and loss estimation analysis Utilized Python, Pandas, Matplotlib, Seaborn, Scikit-Learn, Logistic Regression.

Tools & Technologies: Python, OpenCV, TensorFlow/Keras, U-Net, Image Augmentation, Matplotlib

• Developed a deep learning model for segmenting underwater images to enhance object visibility and boundary detection in murky environments
• Utilized U-Net architecture for semantic segmentation, achieving high pixel-level accuracy on challenging underwater datasets
• Applied image preprocessing techniques (color correction, contrast enhancement) to improve model performance in low-visibility conditions
• Performed data augmentation (rotation, flipping, brightness shift) to increase training dataset robustness
• Evaluated performance using metrics such as IoU (Intersection over Union) and Pixel Accuracy
• Visualized segmentation maps and overlays using Matplotlib and OpenCV for qualitative analysis