Things I have been working on

Projects

Centralized data from multiple sources into various databases using ETL (Alteryx) and SQL.
Designed and implemented data pipelines to automate the extraction, transformation, and loading (ETL) processes.
Utilized Alteryx for data preparation, blending, and analytics, ensuring data integrity and consistency.
Employed SQL queries for efficient data extraction, joining, and aggregation across different databases.
Developed dashboards and reports using Power BI to visualize production metrics and key performance indicators (KPIs).
Integrated data from SAP, MES (Manufacturing Execution Systems), and other operational databases into a centralized data warehouse.
Conducted data quality checks and validation to ensure accurate and reliable data for analysis.
Streamlined data workflows, reducing manual data handling, which resulted in saving up to 1000 human hours.
Provided data-driven insights to optimize production processes, enhance efficiency, and support decision-making.
Collaborated with cross-functional teams, including IT, operations, and finance, to align data strategies with business goals.
Created detailed Power BI dashboards that included interactive visualizations, allowing stakeholders to explore data and derive insights dynamically.
Utilized advanced Power BI features such as DAX (Data Analysis Expressions) and Power Query for complex data transformations and calculations.

Led a team of data scientists to set up a multi-GPU environment on IBM's cluster.
Configured and optimized NVIDIA GPUs for parallel processing to accelerate deep learning model training.
Developed and implemented convolutional neural networks (CNNs) using PyTorch and TensorFlow for vehicle detection from drone images.
Applied transfer learning techniques using pre-trained models such as YOLOv5 and Detectron2 to enhance detection accuracy and reduce training time.
Preprocessed and augmented large datasets of aerial images to improve model robustness and performance.
Utilized IBM Watson for additional cognitive computing capabilities and integration with cloud services.

Created a comprehensive tool to optimize staff mobility in urban areas using advanced algorithms and graph theory.
Developed bio-inspired algorithms, as ant colony optimization, to find optimal routes and schedules for staff transportation.
Implemented the solution in Python, leveraging libraries such as NetworkX for graph-based analysis and Scikit-Learn for machine learning tasks.
Integrated Google Maps API for real-time geolocation data, route planning, and traffic updates to enhance the tool's accuracy and efficiency.
Reduced transportation costs by optimizing travel routes and schedules, resulting in significant savings for the company.

Developed and deployed automation scripts to manage and update the affiliate customer database, enhancing efficiency and reducing manual workload.
Utilized Apache Airflow to create DAGs (Directed Acyclic Graphs) for orchestrating complex data workflows, ensuring smooth and reliable execution of automated tasks.
Established secure FTPS connections for data transfer, ensuring the confidentiality and integrity of sensitive information.
Extracted and processed data from internal APIs, leveraging RESTful services to gather and update information dynamically.
Collaborated closely with backend development teams to integrate automation scripts seamlessly into the existing e-commerce platform.

Developed an automated system for managing an Instagram account, utilizing advanced AI models to generate and curate content.
Leveraged HuggingFace’s models for sentiment analysis , text generation and image generation ensuring relevant and engaging captions for posts.

Created innovative video animations utilizing Deforum and Stable Diffusion for generating frame-by-frame image sequences.
Used Stable Diffusion models to produce high-quality, AI-generated images based on specific prompts and themes.
Integrated Deforum’s animation capabilities to interpolate and transition between frames smoothly, creating fluid and dynamic videos.
Fine-tuned open-source models for specific styles and subjects, enhancing the artistic quality and relevance of the animations.

eveloped a Jupyter notebook for forecasting stock market trends using machine learning techniques.
Leveraged TensorFlow to build and train predictive models based on historical stock data, focusing on time-series analysis.
Utilized LSTM (Long Short-Term Memory) networks to capture the temporal dependencies in stock prices, improving the accuracy of forecasts.
Integrated financial datasets from sources like Yahoo Finance and Alpha Vantage, ensuring comprehensive and up-to-date information for model training.
Visualized forecast results using Matplotlib and Seaborn, providing clear and interpretable charts for stakeholders.

Developed an automated trading script using Alpaca API, EC2, and Python to execute trades based on predefined strategies.
Implemented technical analysis strategies, including moving average crossovers, Bollinger Bands, and momentum indicators, to identify trading signals.
Utilized Alpaca’s commission-free trading API for real-time market data access and order execution.
Deployed the trading script on AWS EC2 to ensure high availability and performance, allowing for 24/7 operation.
Utilized Pandas and NumPy for data manipulation and analysis, ensuring efficient processing of large financial datasets.