vikrantdeshpande09876 / masterize_hospital_entities

The goal was to maintain a ‘single version of truth’ for associated entities across the entire organization’s data sources. The RecordLinkage package is integrated with a wrapper recursive data-pipeline for de-duplicating of records and generating a master set. Similarity between two textual strings determines if they are a probabilistic match.

• 🌱 Always learning and growing as a Data Scientist^meme, and Software Engineer^meme.
• 💬 Ask me about the world of data or astrophysics, and I promise something interesting.
• 👯 Reach out if you're looking to collaborate on Machine Learning side projects, or need a partner for Leetcode problem-solving.
• ✨ Insert random inspirational quote: "The sooner you start to code, the longer the program will take".
• ⚡ I'm not always fun at parties:

programming_languages = [
                  🐍 Python: (Pandas, Numpy, Tensorflow, Scikit, Flask, PySpark, Airflow, BeautifulSoup, etc.),
                  📊 R (ggplot, Tidyverse, etc.),
                  🧮 SQL,
                  ☕ Java (SpringBoot),
                  🐧 Linux Bash Scripting,
                  ｡🇯‌🇸‌ Javascript
]
predictive_models = [LLMs, Prompt Engineering, NLP, Timeseries Forecasting, Classification, Regression, Clustering, Ensembling, Transformers, ...]
statistics_and_exploratory_analytics = [Hypothesis Testing, Power Analyses, Mixed Effect Modeling, Regression Analyses, A/B Testing, ANOVA, ...]
databases = [SQL Server, Postgres, MySQL, MongoDB, ...]
cloud_services = [Azure Machine Learning, Azure Functions, Azure Blob Storage, GCP Cloud Functions, Google Cloud Composer, Google Cloud Storage, AWS S3, RDS, Sagemaker, ...]
tools = [
                  🏷️ Git,
                  🐳 Docker,
                  ☸️ Kubernetes,
                  😊 HuggingFace.
                  🏗️ Tensorflow,
                  Tableau,
                  Heroku,
                  Kafka,
                  Airflow,
                  Informatica Workflows,
                  Jira,
                  Bitbucket,
                  Postman,
                  JMeter,
                  ...
]
cloud_certifications = [
                  Azure Certified Data Scientist,
                  AWS Cloud Certified Practitioner,
                  Deep Learning Specialization,
                  ...
]

• A Microsoft Teams Chatbot with a highly scalable backend for 90+ DAU using Azure OpenAI GPT4, HuggingFace gte4, intent-detection, advanced Retrieval Augmented Generation, and hybrid-search on a vector store (internal company project).
• An NLP-based package to recommend cell-type annotations and help establish a controlled vocabulary in scRNA-seq datasets, using HuggingFace sentence-transformer models.
• A multiple timeseries MLOps system for demand forecasting using ARIMA and FB Prophet models, and exogenous variables like discounts, price-hikes, number of housing-permits, consumer sentiment index, etc. (internship project).
• A Credit Card fraud detection Random Forest based system deployed on GCP to monitor source-file changes using Cloud Functions and Airflow (Cloud Composer).
• A distributed weather-reporting system deployed using microservices architecture in Kubernetes clusters for real-time streaming of inference-reports over Kafka topics.
• An in-house master-data-management tool to recursively parse and merge subsets of data while tracking transitive dependencies, using PySpark, SQL Views, and Levenshtein similarity for string-comparisons.
• An AI product to detect weapons in CCTV/webcam footage and immediately notify authorities using YOLOv3, Docker, Kubernetes, and Kafka.
• Bronze medal for Revenue prediction via a Stacked ensemble-model of Gradient Boosting methods.