Hi, I'm Priya 👋

Let's build something awesome!

I'm an enthusiastic problem-solver and user-centric developer. My professional journey, which spans over 7 years, has taken me through various roles in engineering and operations, from dynamic startups to established corporations. My goal? Crafting solutions that aren't just smart, but also scalable and efficient.

Right now, I'm on the lookout for fresh challenges, hoping to bring my mix of technical know-how and big-picture thinking into a space that's all about pioneering and growth.

So, here's my invitation for you to dive into my portfolio. I'm excited for you to see how my passion and adaptability can really make an impact. Let's embark on this journey of discovery together!

Skills & Qualifications

  • Writing clear, readable code
  • Cross-functional collaboration
  • Constantly learning new skills & techniques
  • Optimizing internal processes
  • Googling the sh$t out of everything
  • Clear & empathetic communication

Education

  • Zero to Mastery Academy
  • Complete Python Developer
  • Complete Machine Learning and Data Science
  • Complete Web Developer

  • University of Pennsylvania
  • Computational Thinking for Problem Solving

  • San Jose State University
  • MFA in Creative Writing
  • BA in English

Recent Projects

Step into the realm of practical applications with my latest work.



Snoop, the Dog Breed Checker

Purpose of the Project

This project uses transfer learning and TensorFlow to train a machine learning model on identifying a dog's breed from an uploaded image of the dog.

Technologies Used

Python, Jupyter Notebooks & Google Colab, Transfer learning using TensorFlow & TensorFlow Hub, Keras, NumPy, Pandas, Scikit-Learn, Matplotlib, Seaborn

Obstacles & Challenges

As of now, the biggest obstacle has been ensuring that I save my models immediately. Since the dataset for this project was relatively large, it took over an hour to train on the full data. Unfortunately, my connection was interrupted soon after this step completed. If I had included a "save model" step within my training function, that may have saved me some time in retraining the model.

Potential Improvements
  • Tune-ups: Further tune the model to reduce possibility of overfitting, or try training on a different multi-classification model from TensorFlow Hub.
  • UI: Create a user-friendly UI using Streamlit so users can upload their images and view the dog breed.
  • Code refactoring: Improve or increase my use of functions throughout the project, especially when training a model and saving a copy to Google Drive.

SparkNotes & ChatGPT Walk into a Bar

Purpose of the Project

This project allows users to interact with the trained ChatGPT-3.5-Turbo LLM model by uploading their own files, setting the chunk size and k parameters, and asking questions about the uploaded file in a conversational manner. The chat history is retained as long as the parameters and file remain unchanged.

Technologies Used

Python, Langchain, OpenAI, Streamlit, Vector Embeddings & Vector Store, Chroma / Pinecone

Obstacles & Challenges

The biggest challenge occurred after deploying the app on Streamlit. Although the development version worked without any issues, the deployed version has been throwing an error related to the pysqlite3 package version, which I'm currently debugging. Another issue is that an OpenAI API key is required to run the app, which involves the user (or myself) paying for each chat conversation.

Potential Improvements
  • Package maintenance: There will need to be a better way to maintain packages in an appropriate version for Streamlit and Python.
  • Additional parameters: Users could select which ChatGPT version they want to use as a dropdown menu.
  • Chat history: Add the option to save or copy the chat history text; also add a warning that the history will disappear once the parameters or file is changed.

Let's Get to Know Each Other

Thank you for exploring my portfolio! I'm always open to discussing new opportunities, innovative projects, or anything else you'd like to talk about.

You can find me on the following platforms:

I'm based in the Bay Area, and open to local or remote opportunities.
Looking forward to the possibility of connecting!