smithla02 / smithla02

Profile README

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Hi there, I'm Luis Smith πŸ‘‹

About Me

I'm a Data Scientist with a strong background in machine learning engineering, specializing in recommendation systems and pricing models.

πŸŽ“ I recently completed a MSc in Computer Science (Machine Learning) at University of Texas at Austin and I'm looking forward to applying machine learning techniques in future projects.

πŸ“« How to reach me: Email | LinkedIn

Skills

  • Programming Languages: Python, SQL, Go, Bash
  • Frameworks: Pandas, NumPy, Scikit-learn, Statsmodels, TensorFlow, Keras, PyTorch, Hugging Face, Spark, Docker, GitLab, Google Cloud Platform (GCP), Google BigQuery, dbt, Streamlit, Apache Airflow, Apache Kafka, Apache Flink, Tableau, Redis, Git, Excel

Projects

  • Research Projects

    • 3 research projects conducted during my masters
      • Improving Lexical Reasoning of NLI Models

        • Enhanced Natural Language Inference (NLI) models' ability to discern lexical relationships using data augmentation and adversarial training
        • Technologies: Hugging Face (Transformers), Pandas
      • AI Agent for Ice Hockey in SuperTuxKart

        • Developed an image-based AI agent with hand-tuned controllers for playing ice hockey in SuperTuxKart, leveraging Fully Convolutional Networks (FCN)
        • Technologies: PyTorch
      • Real Estate Investment Recommendations

        • Explored machine learning and financial models to provide real estate investment recommendations
        • Technologies: ARIMA, Prophet, PyTorch
  • Side Projects

    Feel free to check out the repositories for more detailed information on each project.

Interviews

Education

  • MSc in Computer Science (Machine Learning), University of Texas at Austin
  • BA in Economics, Stanford University

Work Experience

  • Specialized in Recommendation Systems, Contextual Bandits, and Linear Programming.
  • Senior Data Scientist, Pricing, Gojek (Jan 2020 - Sep 2021)
    • Engineered data and model training pipeline for TensorFlow-based pricing algorithm and linear programming-based price reconciliation, leading to 10% performance uplift in gross profit
    • Developed ML pricing web service in Go that performed experimentation, feature retrieval, model prediction, and price presentation with < 50 ms SLA
    • Applied linear programming techniques to translate government regulations and business constraints into linear constraints within pricing models
  • Lead Data Scientist, Recommendations, Gojek (April 2018 - Jan 2020)
    • Led a team of 3 data scientists and data analysts that optimized a food delivery recommendation system
    • Built recommendation model driving 2 million incremental bookings every month and leading to 15% performance uplift versus existing models (Patent No. 10201912472U).

Interests

Traveling, hiking, photography, electronic music, meditation.

About

Profile README