Juan Acosta Portfolio

About Me

Hello! I'm Juan Acosta, a data scientist who loves turning data into meaningful insights. My journey from a music teacher to a tech enthusiast has equipped me with strong analytical and collaborative skills. I specialize in data acquisition, modeling, machine learning, and NLP, with a keen interest in urban planning and transportation. Along the way, I've picked up some solid coding chops that keep my tech toolkit sharp.

When I'm not diving into data, you'll find me immersed in jazz, reminiscing about my drumming days, watching soccer, enjoying a good cup of coffee, or strategizing my next move in a chess game. Passionate about using data for positive change and with a professional interest in AI ethics, I'm eager to connect and make a difference together!

Skills

Python
SQL
- MySQL
- SQLite
Machine Learning
- Scikit-Learn
- TensorFlow
- Keras
- XGBoost
- PyTorch
Data Analysis
- Pandas
- NumPy
- Jupyter
A/B Testing
- SciPy
AWS
- EC2
- S3
- Lambda
Data Visualization
- Matplotlib
- Seaborn
- Plotly
- Tableau
Web Scraping
- BeautifulSoup
APIs
- RESTful APIs
- Postman
- Flask
- Google Maps API
Git & GitHub
- Git
- GitHub
NLP
- NLTK
- BERT
Audio Processing
- Librosa

Projects

Real Estate Chatbot

A WhatsApp chatbot for real estate inquiries, built using Twilio, Flask, and BERT. It uses LangChain and RAG for managing context and delivering accurate property information, automating client interactions while easing the workload for agents.

Urban Sound Classifier

Classify different urban sounds using machine learning techniques. Features extracted from audio signals, tested multiple ML models for prediction, with XGBoost achieving ~70% accuracy.

Speech Recognition

Create a neural network to recognize voices of 10 different individuals. Extracted and preprocessed Mel-Spectrograms and MFCCs, tested different neural networks, with Bidirectional LSTM layers yielding 85% accuracy.

Distance Matrix

Create distance matrices and retrieve information from Google Maps API. Set up an API to retrieve information quickly, format it for Python and Pandas, and store it properly with functionality for SQL integration.

Blog

Building a Real Estate WhatsApp Chatbot with AI and Automation

Discover how I built a fully automated WhatsApp chatbot for real estate inquiries, using cutting-edge tools like Twilio, Flask, BERT, and LangChain. Learn how AI-driven technologies streamline client interactions and reduce real estate agent workloads, with a step-by-step breakdown of the process. Perfect for developers, data scientists, and anyone curious about chatbot development.

Data-Driven Urban Planning: Illuminating City Landscapes through Polar Histograms and OSMnx

The blog post explores the application of OSMnx and polar histograms in urban informatics to visualize and analyze city structures and transportation networks. This approach enhances the understanding of urban environments and supports the development of smarter, more sustainable cities.

Using Google Maps Distance Matrix API to Create a Distance Table

This post demonstrates how to call the Google Maps API and create a distance matrix using Pandas. It also covers storing data using SQLite for efficient retrieval and manipulation.

Unveiling Karatsuba Multiplication: A Breakthrough in Algorithmic Efficiency

This post breaks down the Karatsuba multiplication algorithm mathematically and provides Python code for its implementation, showcasing its efficiency in computational mathematics.

Education

Flatiron School

Program: Immersive Data Science Bootcamp

Location: New York, NY

Description: Completed an intensive data science bootcamp focused on data acquisition, data modeling, statistical analysis, machine learning, deep learning, and natural language processing (NLP).

SUNY Purchase

Degree: Master of Music in Jazz Performance

Location: Purchase, NY

Webster University

Degree: Bachelor of Music in Jazz Studies

Location: St. Louis, MO

Juan Acosta

Data Scientist

NYC