Rakhi Kumari
AI Engineer
******.*******@*****.*** 083********
linkedin.com/in/rakhi-kumari-ab521769
github.com/kumarirakhi19
PROFILE
Data scientist professional with 5+ years of experience specializing in data analytics and machine learning working in different sector. Proficient in SQL,, LLM Models, Gen AI and NLP with a strong ability to extract valuable insights from data. Known for being a quick learner, efficient team player, and possessing excellent communication skills.
SKILLS
Programming Language, Database & ML Libraries
Python, SQL, NumPy, Pandas, Scikit-learn, Matplotlib, Polty, Seaborn, Tensorflow, Tableau, R,PyOCR Programming, Big query,Gitlab,cloud storage, pytorch, Apache Spark
DATABASES
PostgresSQL, MySQL,SSD
Algorithms
Gen AI, Mlops, CI/CD Pipeline, RAG, Fine-Tuning, Feature Engineering, XGBoost, NLP, open CV, Hypothesis testing, Tensorflow, Keras, CNN, WebScraping, Yolo, BERT
Platforms
Power BI, Tableau, MS Excel, MySQL, PostgreSQL, Docker, Flask, MLFlow,Git, Advanced Excel, AWS(S3, Coud watch,lambda, aws bedrock,EC2),FastAPI
COURSES
Full stack data science (2021)
R programming and R Shiny Dashboard (US Arrest and Star Wars)
AWS Sagemaker
Power BI(Simplilearn)
Internship(Data science)-BELONG
EDUCATION
M.SC Data Science
Kalasalingam University
M.B.A
Nagpur University
B.SC(Interdiciplinary)
Nagpur University
PUBLICATIONS
Data science in Aerospace
Medium
INTERESTS
cooking Badminton Reading
PROFESSIONAL EXPERIENCE
AI Engineer
Mea Platform(01/2023 - Present)
Achievements/Tasks
•Primary Responsibility is Revenue Generation and Strategy making with clients. Communicate client's business requirement constructing easy-to- understand data and deliver to department team. Ability to manage time of clients to repeated business with clients
•Implemented OCR techniques to extract text from images and scanned documents, Layout Analysis, data extraction .
•Involved in Open CV, LLAMA, BERT with fine- tunning and Preprocessing,NLP, text extraction, summarization, transcription, chatbot development with RAG, Generative AI with AWS Bedrock.
•Annotated the data set on Azure platform. Performed weekly data analysis for data extraction and transformation, cleaning, summarization on excel and creating visualization dashboard that provided real-time KPIs.
Data Scientist
Almabetter pvt ltd
12/2021 – 12/2022
•Constructed linear forest for R & D datasets with 200 records or fewer.Evaluated the performance of a regression model along with std deviation and errors.
•Designed LOOCV to help with insufficient data issues, & improve performance by 20%
•Executed POC of linear forest to analyze the performance of multiple datasets.
Learning and development Executive
Pushpanjali Hospital & Medivic Pharmaceutical Pvt Ltd 07/2018 – 12/2020
•Applied data science model to hospital data to identify patterns in patient visits, medicine usage, and staff efficiency. These insights supported data-driven decisions to optimize operations and improve healthcare service delivery.
PROJECTS
Gemini Pro LLM MultiplePDF Chat - Generative AI Model 02/2024 – 07/2024
•Developed a web application using Streamlit and Google’s Generative AI model Gemini to enable chat-based interaction with multiple PDF files. Users can upload PDFs, ask questions related to the content and receive answers generated by Gemini.
•Utilized FAISS for vector embedding, RAG and Langchain library for natural language processing tasks, questions-answering. Employed PyPDF2 library for text extraction from PDF files.
•Streamlined the deployment process, reducing manual intervention and minimizing deployment errors.
Insurance entity extraction
•Implemented data ingestion with AWS Lambda and S3.
•Utilized Amazon Textract for OCR and a sliding window matching algorithm for accurate entity extraction.
•Integrated Anthropic Claude 3.5 Sonnet via AWS Bedrock, cloud optimisation using PDFs and custom prompts for natural language understanding and extraction.
Brain Tumour Classification and segmentation
•Unet, unet++ architecture with dense skip connection for multi scale feature learning.
•Extended Unet with attention gates to focus on tumor regions.
•Implemented Vit model for classification of tumor brain with 97 % accuracy .
Nlp2sql
•Developed an Insurance AI Analytics using LangGraph and Amazon Bedrock (Claude) to convert natural language insurance queries into SQL for underwriting and claims analytics.