• Built scalable ML pipelines by automating data ingestion and transformation using REST APIs and AWS EC2, following MLOps best practices.
• Deployed LSTM-based models (Vanilla, Bi-LSTM) to forecast energy consumption, improving prediction accuracy by 60% and enhancing business demand planning.
• Designed a normalized PostgreSQL schema for time-series data, accelerating query performance by 90% and reducing dashboard load times.
• Improved production workflows by integrating model serving and versioning systems, enabling faster and more reliable model updates.
• Tools: Python, Flask, TensorFlow, PostgreSQL, AWS EC2

• Automated data collection pipelines using Selenium, BeautifulSoup, and cron jobs, reducing manual extraction efforts by 70% and ensuring real-time offshore construction metric tracking.
• Designed interactive Power BI dashboards to visualize wave sensor data, enhancing decision-making for offshore wind project safety and future strategy planning.
• Fine-tuned LLaMA 3 models on scraped scientific documents with Hugging Face Transformers, building an agentic RAG-based pipeline for semantic knowledge retrieval and open-ended scientific inquiry simulation.
• Tools: Python, Selenium, BeautifulSoup, Power BI, Hugging Face Transformers, LLaMA 3

• Analyzed large-scale insurance records using PySpark on Azure Databricks by connecting to Azure Data Lake, applying Apriori and FP-Growth algorithms to detect fraud and risk patterns.
• Developed an R-CNN-based model to classify diverse insurance claims documents, improving triage accuracy and reducing manual processing time, enhancing claim authenticity validation by 36%.
• Built interactive dashboards in Power BI to visualize fraud risk patterns, enabling stakeholders to monitor high-risk claims, prioritize investigations, and drive data-driven decisions in fraud management.
• Tools: PySpark, Azure Databricks, Azure Data Lake, R-CNN, Power BI

• Developed and tested multiple NSE option trading strategies using Zerodha API, formatting data for easy access and robust live market validation.
• Applied technical indicators (VWAP, RSI, MA, MACD) to trigger automated position entries and exits in real-time trading environments.
• Monitored and optimized financial indicators to enhance ROI by 50% while implementing dynamic risk management strategies to mitigate stop-loss risks.
• Tools: Python, Zerodha API, Technical Analysis

• Directed the development of a loan approval workflow by analyzing credit histories, improving the approval process effectiveness by 70%.
• Engineered features using VIF and RFE techniques and built classification models achieving 90% accuracy.
• Deployed the entire ML pipeline on the cloud using Streamlit and GitHub, ensuring seamless accessibility and real-time collaboration for students.
• Tools: Python, Streamlit, GitHub, Machine Learning (VIF, RFE)

AI / Machine Learning:

• TensorFlow, PyTorch, Scikit-learn, Keras, XGBoost
• Hugging Face Transformers, LLaMA, RAG Pipelines
• Deep Learning, Computer Vision, NLP

Cloud Platforms & MLOps:

• AWS EC2, GCP, Azure
• Docker, Kubernetes (basic), MLflow
• Streamlit, FastAPI, Heroku

Databases:

• MySQL, PostgreSQL, MongoDB
• Azure Data Lake, Snowflake

Programming Languages:

• Python, Java, JavaScript, C, C++, Bash, HTML5, CSS3

Libraries & Tools:

• NumPy, Pandas, OpenCV, Matplotlib, SciPy, Dask
• Git, Power BI, Tableau, Excel