• Significantly enhanced seed forecasting accuracy by 25% through the successful creation and deployment of a Machine Learning model, leveraging weather data to drive agricultural decision-making.
• Ensured robust data integrity and system reliability by implementing 2 monitoring and alerting frameworks within the pipeline, contributing to the seamless operation of the technology stack comprising Dataflow, BigQuery, Cloud Functions, Stack Driver, Cloud Scheduler, Cloud Storage, Monitoring & Alerting, Cloud Run, and XG Boost.
• Successfully led a team of three data scientists in optimizing various ML parameters, resulting in a notable improvement in model performance, demonstrating effective leadership in the intersection of agriculture, technology, and data science.

• Boosted retail order forecasting accuracy by 30% across all MMS outlets through the development of two ML pipelines, emphasizing precision in demand prediction for the retail sector.
• Analyzed sales trends during key events (Black Friday, Easter, Christmas, and Corona) with 3 implemented BigQuery models, showcasing strategic data insights.
• Streamlined on-premises to BigQuery data migration with 3 CI/CD stream pipelines, reflecting a commitment to efficient and agile data management.
• Achieved a 15% cost reduction through the creation of 5 operational dashboards for multi-country business processes, leveraging a diverse tech stack, including Pub-Sub, FTP, SFTP, App Engine, Compute Engine, Looker Studio, Jupyter Notebooks, Jira, Git, Slack, and Cloud IAM.

• Developed full-stack neural networks using Keras for Lesara.com, using 6 datasets to accurately predict the gender and amount of visitors.
• Predicted rent prices for luxury apartments in Texas, USA using KNN and Decision Trees.
• Developed ML model to predict bike prices for Berlin shops using 50 years of data, improving efficiency by 29%.
• Tech Stack - Scikit-learn, Pandas, Matplotlib, Seaborn, GridSerchCV, Tensorow, Keras.

• Built a Widget in Orange (Open Source ML) to parallelize data using MPI (Python).
• GUI Library Application for storing & updating books using Tkinter & PostgreSQL.
• Astronomy Application using MYSQL to nd the position of planets from a location.
• Web Scrapping to parse prices & location from a Real Estate Website using BeautifulSoup4.
• Web Map Application using Folium to arrange places according to Volcano eruption.

Master Thesis - Scalable Architecture for Entity Recognition and Disambiguation

• Develop a scalable full-stack web application that extracts person and location names from texts and disambiguates them from Wikipedia to find probable real-world Wikipedia entries.
• Tech Stack - NumPy, Cosine Similarity, Natural Language Toolkit (NLTK), Stanford Named Entity Recogniser, Count Vectorizer, Punkt Tokenizer, German Stemmer, TF-IDF.