Hi 👋,
My name is Kaustubh Bhavsar, and I hold a Master of Technology degree in Machine Learning and Artificial Intelligence from Lovely Professional University, India. During my pursuit of this degree, I had the opportunity to conduct research at the intersection of healthcare and artificial intelligence, with a specific focus on improving patient care and enhancing healthcare processes through the optimization of medical diagnostics. Additionally, my study involved a thorough exploration of how AI is revolutionizing the healthcare industry, paving the way for innovative solutions and transformative advancements.
I regularly engage in composing concise blog entries on Twitter pertaining to machine learning and associated concepts. Please take a look at the compilation at the designated repository to explore a captivating collection of my most relevant tweets: Repository Link.
Journal Publications (Google Scholar):
-
A Comprehensive Review on Medical Diagnosis Using Machine Learning [38 citations, Nov'23]
Kaustubh Arun Bhavsar, Ahed Abugabah, Jimmy Singla, Ahmad Ali AlZubi, Ali Kashif Bashir, Nikita
Computers, Materials & Continua (CMC), 2021, 67(2), 1997-2014 -
Medical Diagnosis Using Machine Learning: A Statistical Review [35 citations, Nov'23]
Kaustubh Arun Bhavsar, Jimmy Singla, Yasser D. Al-Otaibi, Oh-Young Song, Yousaf Bin Zikria, Ali Kashif Bashir
Computers, Materials & Continua (CMC), 2021, 67(1), 107-125
Blog Posts (Medium):
- Prompt Engineering for Arithmetic Reasoning Problems: Explored various prompt engineering techniques for arithmetic reasoning problems, best practices, and rapid experimentations for production-grade prompts with Vellum.ai.
- Stemming: Porter Vs. Snowball Vs. Lancaster: Described how various stemmers operate and the distinction between stemming and lemmatization.
Relevant Selected Projects:
FastRide: NYC Taxi Trip Duration Prediction
Machine Learning based application that utilizes Weights&Biases for experiment tracking, Hopsworks Feature Store for data storage, and Alibi-Detect for monitoring, for accurately predicting the duration of taxi trips in New York City, providing valuable insights for businesses in the transportation industry to optimize operations and improve customer satisfaction, while employing a serverless architecture and featuring a Streamlit web-app. Tools & Libraries: Weights&Biases, Hopsworks (feature store), Alibi-Detect, Great-Expectations, Pytest, Streamlit, Folium, Scikit-Learn, Xgboost, Pandas, Numpy, Matplotlib, Seaborn, Yellowbrick. |
|
SliceMate
GPT-3.5 based information extractor that intelligently extracts, stores, and retrieves personal information from natural language inputs, ranging from shopping lists to random ideas, facilitating easy organization and discoverability of diverse data. Tools & Libraries: OpenAI, Streamlit, Pandas |
|
Intentium
Multiclass intent classification (including subtask: topic modelling) using various models such as MLP, LSTM, and BERT, with comparison analysis and summarization of results, and best model served via FastAPI. Tools & Libraries: Tensorflow, Keras, Tensorflow-Hub, Scikit-Learn, Gensim, NLTK, Spacy, PyLDAviz, FastAPI, Re, WordCloud, Matplotlib, Seaborn, Numpy, Pandas. |
|
Animalica
Multiclass (10 class) image classification system (26,000+ data points) that addresses class imbalance through the application of class weights, incorporates transfer learning techniques, and deploys the optimal model via a Flask API. Tools & Libraries: Tensorflow, Keras, Scikit-Learn, Matplotlib, Seaborn, Numpy, Pandas, Flask. |
|
Model Comparison Workflow on Issue Pull: Weights & Biases
Continuous Integration/Continuous Deployment (CI/CD) workflow implemented using GitHub Actions and the Weights&Biases platform, which automatically generates a model comparison report when a user creates an issue on GitHub with a specific command |
|
Sentiment Analysis
Comparative analysis of various encoding schemes of Bag-of-Words for Sentiment Analysis using Neural Network: Determining the worthiness of a movie. Tools & Libraries: NLTK, Keras, Tokenizer, WordCloud, Re, Matplotlib, Seaborn, Numpy, Pandas, String. |
Select Courses/Certifications:
- Serverless Machine Learning (by Jim Dowling, CEO of Hopsworks & Associate Professor at KTH, Sweden)
- Practical Data Science on the AWS Cloud (Deeplearning.ai - Coursera)
- Effective MLOps - CI/CD for ML (Weights & Biases)
- Machine Learning Specialization (Stanford Online - Coursera)
Collaboration and Contact Information:
I'm open to collaborating on research, application building, content writing, moderation, or competitions in NLP and ML.
How to reach me: Shoot an email to [email protected] or through my social media profiles -
- Twitter: twitter.com/bhavsarkaustubh
- Linkedin: linkedin.com/in/kaustubhbhavsar