Aditi Nishankar

Projects:-

Title: User Comments Summarization

(NLP, NER, Spacy, BART, Python, PyTorch)

• Developed an LLAMA-based text summarization model to streamline OEM comment processing for inventory delivery waivers, improving warehouse efficiency.
• Created an API to manage and update summarized comments, optimizing the waiver process and enhancing charge management.
• Generated $10 million annually in value for stakeholders.

Title: Accessory Recommender System.

• Built a recommendation engine suggesting three accessories based on the relationship with phones, and location-specific inventory levels.
• Developed a weekly accessory bundle recommendation system, increasing sales by 25%.

Title: Chatbot.

(Hugging Face, LangChain, LLM, ChromaDB, Python)

Key responsibilities:

• Set up a local architecture for the LLAMA model as an API service, integrating it with applications to power an intelligent chatbot that enhances data retrieval and analysis.
• Developed and deployed a responsive chatbot based on LLAMA, improving business operations by providing accurate information quickly.
• Currently enhancing the system to query dashboard data and deliver role-based insights to stakeholders.

Title:- Time Series Forecasting for Recovery Data.

Key Responsibilities:

(ARIMA, FB PROPHET, Python, SQL)

• Conducted in-depth data analysis to unveil patterns and
  anomalies in historical recovery data.
• Achieved an impressive 92% accuracy rate by fine-tuning models and optimizing hyper-parameters.
• Continuously monitored and updated forecasting models to adapt to evolving recovery trends, ensuring long-term accuracy.

• Developed and maintained Python scripts to automate routine tasks.
• Played a key role in creating high-quality annotated datasets for training computer vision models, focusing on image segmentation projects.
• Expanded knowledge of Python libraries and frameworks,
  including NumPy, and Pandas through practical application in real-world projects.