Skip to content
View Praful932's full-sized avatar
🦾
tinkering
🦾
tinkering

Block or report Praful932

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
praful932/README.md
########  ########     ###    ######## ##     ## ##       
##     ## ##     ##   ## ##   ##       ##     ## ##       
##     ## ##     ##  ##   ##  ##       ##     ## ##       
########  ########  ##     ## ######   ##     ## ##       
##        ##   ##   ######### ##       ##     ## ##       
##        ##    ##  ##     ## ##       ##     ## ##       
##        ##     ## ##     ## ##        #######  ######## 

Senior ML Engineer → ML Researcher (in progress)
AI Governance @ Visa · LLM Internals · SLMs · NLP


👨‍💻 About Me

  • 🔬 Senior ML Engineer @ Visa — building AI Observatory, an internal platform to monitor & govern traditional ML models, GenAI apps, and agents across Visa
  • 📐 Building toward a ML Research career
  • 🎯 Core interests: decoding strategies, explainability in LLMs, small language models, ML in healthcare
  • ✍️ Break things down at praful932.com
  • 🕹️ Fun fact: I ran a game zone on my own computer when I was 12
  • hmu

🔭 What I'm Working On

current_focus = {
    "day_job":     "AI Observatory — ML governance platform @ Visa",
    "interests":   ["SLMs", "decoding", "LLM explainability", "ML x healthcare"],
    "hobbies": ["anime", "ukulele🎸", "books", "travel 🏖️", "calisthenics"]
}
 
past = {
    "Prodigal (YC)": [
        "ML engine for ProNotes — ~100k call summaries/day",
        "finetuned LLMs (Llama, proprietary) for call understanding",
        "NER model for healthcare PII redaction — +9% F1",
        "human-connect classifier — +11% recall",
    ],
    "stack": ["PyTorch", "AWS SageMaker", "Databricks", "Llama", "OpenAI"],
}


🧐 Some of My Work

  • antibody-developability-abmelt — Antibody developability benchmark in collaboration with HuggingFace Science.
  • llmsearch — Find better generation parameters for your LLM.
  • Tf-Rec — Python package for building recommender systems on top of Keras/TF2 with GPU acceleration. Published to PyPI, with docs.
  • Kitabe — Book recommendation system. Rate books, get instant recommendations.
  • Machine-Learning-Implementations — ML algorithms from scratch: linear regression, neural networks, SVMs, PCA, collaborative filtering.
  • Edutopia — Platform to connect students with mentors. Final project for CS50. Live here.
  • EverythingEV — Electric vehicle ecosystem web app built for Smart India Hackathon 2020 (VMware problem statement). Live here.

Let's Connect 😎

Pinned Loading

  1. Kitabe Kitabe Public

    Book Recommendation System built for Book Lovers📖. Simply Rate ⭐ some books and get immediate recommendations🤩

    HTML 180 156

  2. llmsearch llmsearch Public

    Find better generation parameters for your LLM

    Python 27 1

  3. blog blog Public

    my personal blog

    JavaScript 7

  4. Tf-Rec Tf-Rec Public

    Tf-Rec is a python💻 package for building⚒ Recommender Systems. It is built on top of Keras and Tensorflow 2 to utilize GPU Acceleration during training.

    Python 18 2

  5. Machine-Learning-Implementations Machine-Learning-Implementations Public

    Repository contains Implementations of Popular Machine Learning Algorithms from scratch in Python

    Jupyter Notebook 7 3

  6. antibody-developability-abmelt antibody-developability-abmelt Public

    abmelt benchmark

    Python 3 5