Raj Doshi Email & Phone Number

New York, Ny, Us

New York, United States

Deeply engaged in leveraging LLM'S to drive impactful outcomes within the organization and utilizing RAG in Knowledge Graphs (Neo4j) to perform tasks and Graph-Based Recommendations. Additionally, extracting valuable insights for CRM optimization through Product Extractions, thereby enabling the team to better understand and serve customers. Involved in developing and deploying innovative solutions, and unvetted Advisor Flow Pipeline which is aimed at enhancing operational efficiency and expediting decision-making processes.

Whippany, New Jersey, United States

1. Fine-tuned Retriever-Augmented Generator (RAG), a LLM using Facebook/DPR retriever on OMOP data to convert Text 2 SQL queries resulting in a significant improvement in efficiency and accuracy.2. Leveraged FAISS to improve search efficiency and aid in MIPS (maximum inner product search) for retrieving nearest neighbors, resulting in a 20% reduction in retrieval time and faster SQL query processing.3. Designed and trained the model utilizing AWS Sage Maker that achieved a high exact match score of 74.7% and deployed the solution on DEEP-MIND to empower Bayer in generating SQL queries and providing valuable assistance.4. Empowered Semantic Scholar and PMC (PubMed Central) API to extract top 10 relevant academic papers, allowing access to a massive database of 200M+ academic papers.5. Leveraged "Refine" Chain technique in LangChain and integrated QA over doc agent, drastically enhancing summarization efficiency by resolving ChatGPT "max limit" issue of 4k tokens and enabling smooth processing of large-scale documents.6. Fine-tuned BioLinkBERT-large model for downstream token-classification NER task in the biomedical domain, successfully recognizing and classifying 84 distinct bio-medical entities.7. Conducted extensive data pre-processing, and text augmentation through GPT- 4 to convert the 400 MACCROBAT data documents, into IOB-2 format resulting in an F1-score of 86% surpassing the benchmark by 8%.

Boston, Massachusetts, United States

1. Monitoring 35+ residence hall access for 400+ residents and their guests at proctor stations.2. Managing emergency situations such as fire alarms, intruder alarms and providing medical assistance.

Teaching Assistant for the course "Data mining Techniques” (CS6200) at Northeastern University.

Palo Alto, California, United States

1. Led 6 Data science interns and more than 40 students in an Agile Format to develop a fine-tuned Open AI GPT-2 124M transformer model to build eclectic AI Text-generation products based on random prompts.2. Developed an end-to-end web application based on Flask by creating a Discord bot and leveraging web scraping methods with Beautiful Soup and SerpApi to produce Netflix-descriptions, code generation and image detection models.3. Devised the NLP curriculum and taught NLP, Machine Learning, Deep Learning, and Flask from scratch, as well as the deployment of the products, which were highly trained for 4+ hours utilizing GPU on the CoCalc server.

Mumbai, Maharashtra, India

●Designed RESTful APIs in Flask, ensuring seamless communication of 5+ components of an E-Learning online platform.●Created extensive dashboards with real-time visualizations, leading to a 30% increase in decision-making for stakeholders.●Led collaborative deployment on AWS cloud, capitalizing Elastic Beanstalk enhanced scalability, supporting 100+ students.

Mumbai, Maharashtra, India

1. Developed an online marketplace for buying and selling goods of more than 300+ products, where users can post their ads which would be delineated to the registered users and find buyers in a hassle-free way.2. Worked on REST APIs in Flask to create the admin page, new user registration, ad portal, and designed many screens for the user interface utilizing HTML, CSS, and Bootstrap to meet business needs.

Master Of Science - Ms, Data Science, 4.0/4.0

Relevant Course Work:- Fall 2021 - CS5800 Algorithms [A] - DS5220 Supervised Machine Learning and Learning Theory [A] Spring 2022 - DS5110.

Bachelor Of Technology, Computer Science, 3.75/4

Computer Science, 84%

Ssc, 95.6%