Machine Learning Engineer

AI/ML Development Cambridge, Massachusetts, United States

Profile
Contacts

Summary

Improved LLM Factual Grounding with Retrieval-Augmented Generation
(Generative AI - Hugging Face - LangChain - Pinecone) Apr 2024 - May 2024
Enhanced the capabilities of a large language model (LLM) by developing a Retrieval-Augmented Generation (RAG) system, leveraging Hugging Face Transformers and LangChain.
Integrated Llama-13b-chat, a large language model from Hugging Face Transformers, for text generation, and implemented a Pinecone vector store for efficient document retrieval based on pre-trained sentence embeddings.
Leveraged LangChain libraries (HuggingFacePipeline and PineconeVectorStore) to construct the RAG system, enabling seamless LLM integration and efficient document retrieval, ultimately enhancing the factual grounding and response accuracy of the LLM.
Data-Driven Product Recommendation System
(Databricks Python SQL - Kaggle) Jan 2024 - Feb 2024
Developed and deployed a robust ETL pipeline within Databricks, meticulously engineered to cleanse and structure an extensive dataset of approximately 1.4 million records extracted from an Amazon product category Kaggle repository.
Leveraged data visualization techniques, including correlation matrices, bar charts, and scatter plots, to extract valuable insights and uncover trends in e-commerce data.
Utilized Logistic Regression Model to predict the best-selling products in the Amazon product category by analysing e-commerce features like price, rating, and reviews. The model achieved an accuracy of ~70% enabling data-driven product identification for potential high sales volume.
Cloud-Based Text Sentiment Analysis
(Python TensorFlow Flask Google Cloud) Nov 2023 - Jan 2024
Developed and deployed a cloud-based text sentiment analysis web application on Google Cloud Platform (GCP) using Flask, achieving an accuracy of ~85% on sentiment classification with a pre-trained model.
Leveraged Google Cloud Storage for scalable and secure management of pre-trained sentiment analysis model and tokenizer.

Expectations

I am looking for a technical role in the field of AI. Where I can get hands on experience with training models and fine tuning them as per the business need.