Chatbots using Generative AI, LLMs, openAI, python, NLP samples

fastAPI LLM, Langchain, Pinecone, openAI in AWS EC2, conversational
* please note Pinecone deletes the indexes if they are not used but am building a chron job + selenium so it’s used at least once every 2 days

Interact with Live Website Data Stored in Pinecone

Live Apphttp://54.198.111.198:8080/static/index.html
GitHubhttps://github.com/data-science-nerds/fastapi-langchain
Premisean original short story of Maria B V who invented
‘qi-fi’, written by Elsa + chatGPT
LanguagesPython, HTML, CSS, and JavaScript
Business/
Use Case
LLM answers interacting with live website content
Showcased technologies
Tech Stack
  1. AI and Machine Learning:
    • langchain: A cornerstone of the AI ecosystem, enabling sophisticated artificial intelligence models and applications.
    • openai, tiktoken: OpenAI’s tools, vital for AI computations and token manipulations.
    • pinecone-client, pinecone_datasets: Machine learning capabilities, encompassing model deployment, management, and dataset manipulations.
  2. Data Analysis and Computation:
    • pandas, numpy, numexpr: These form the backbone of any serious data science operation, from data manipulation to numerical computations.
    • pyarrow: An interface to Apache Arrow, catering to in-memory computing, which is crucial for efficient data processing tasks.
  3. Data Serialization and API Integration:
    • pydantic, marshmallow: Libraries handling data validation, serialization, and deserialization, fundamental when orchestrating ML models with other systems or APIs.
    • protobuf: Enables serialization of structured data – a popular choice in ML and deep learning pipelines.
  4. Web and Asynchronous Programming:
    • fastapi, uvicorn: Modern tools for building and serving APIs efficiently.
    • aiohttp, anyio, async-timeout: Asynchronous libraries ensuring responsive and high-concurrency operations.
  5. Authentication, Security, and Databases:
    • SQLAlchemy: A comprehensive toolkit for database operations.
    • cryptography, rsa, oauthlib: These ensure secure data operations and robust authentication mechanisms.
  6. Logging, Debugging, and Utilities:
    • loguru: A logging utility that can be indispensable when diagnosing issues.
    • tqdm: A progress bar utility, often used in data processing to provide insights into task completions.
Flask, Generative AI using openAI, llama, chatbot limited context

Sample rental lease contract as the only context LLM will answer about

Live Apphttp://3.92.237.8:8000/demo-customized-chatbot
GitHubhttps://github.com/data-science-nerds/ai-architects
PremiseGenerative AI for sample rental lease data, increased cyber-safety in openAI endpoint
LanguagesPython, HTML, CSS, and JavaScript
Business/
Use Case
LLM answers interacting with live website content
Showcased technologies
Tech Stack
  1. AI and Machine Learning:
    • langchain, langchainplus-sdk: Central to the artificial intelligence ecosystem, these are paramount in powering AI models and systems.
    • tiktoken: From OpenAI, for token counting or manipulation, it’s crucial in AI computations.
  2. Data Analysis, Processing, and Science Tools:
    • pandas: An indispensable tool for data manipulation and analysis.
    • numpy, numexpr: Essential libraries for numerical operations, fundamental in data science.
  3. Jupyter and Interactive Computing:
    • Components like jupyter, ipykernel, ipython, and related packages facilitate interactive computing, a staple in data science explorations.
  4. Data Serialization and API Tools:
    • pydantic, marshmallow, jsonschema, fastjsonschema: These handle data validation, serialization, and deserialization, crucial when interfacing ML models with other systems.
  5. Authentication and Security:
    • Libraries like argon2-cffi, cryptography, and rsa ensure robust cryptographic operations, while Flask-Cognito and Flask-Login manage user authentication.
  6. Testing and Debugging:
    • pytest offers comprehensive testing capabilities, and ipdb ensures efficient debugging, both vital for ensuring optimal ML and LLM performance.
  7. Databases and ORM:
    • SQLAlchemy: A renowned SQL toolkit and ORM, essential for data management in projects.
  8. Web Development and Asynchronous Programming:
    • Flask, gunicorn, aiohttp, and related libraries provide a foundation for web development, ensuring responsive and scalable web applications.
Low Code/ No Code WordPress Chatbot, conversational

Small business website that answers questions about the business

Live Apphttps://ai-architects.cloud
GitHubNone
PremiseContinually changing website that answers generic
client questions
LanguagesHTML
Business/
Use Case
Improved small business clients’ experience, increased profits for business owner with fewer employees
Showcased technologies
Tech Stack
  1. Web Development:
    • WordPress: Popular business and blogging website.
    • HTML: Online content.
    • Plugins: WordPress wrappers.
NLP, Business Outcomes Improvement

Post Dop Adoption Retention Rates- what really gets the dogs adopted?

Live Apphttp://3.91.59.182:8000/
GitHubbackend:
https://github.com/elsaVelazquez/faster-pet-adoption-app
frontend:
https://github.com/elsaVelazquez/faster-pet-adoption
PremiseML, NLP, TF/IDF and Cosine Similarity for wrangled API data from animal shelter
Languages
Business/
Use Case
Data Analysis to determine biggest bang for the buck
for dog adoption rates per national data
Showcased technologies
Tech Stack
  1. Natural Language Processing and Data Analysis:
    • nltk, textblob: These are powerhouses in the world of natural language processing, essential for textual data analysis and manipulation.
    • pandas, numpy: Renowned libraries in the data science arena, invaluable for data manipulation, analysis, and numerical computations.
    • scikit-learn, scipy: Key players in machine learning and scientific computations for sophisticated data modeling and analysis capabilities.
    • joblib, tqdm: Utilities that enhance efficiency in data operations and provide insights into computational tasks.
  2. Web Development and Application Deployment:
    • Flask, Werkzeug: Flask, supported by Werkzeug, is a go-to for developing lightweight web applications.
    • gunicorn: A popular WSGI server, not just for developing applications, but also deploying them in robust environments.
  3. Data Collection and Integration:
    • requests, beautifulsoup4: Pivotal for web scraping, data collection, and API interactions, to integrate diverse data sources seamlessly.
  4. Interactive Computing and Development:
    • ipython, ipykernel, jupyter-core: Components that cater to interactive computing, allowing for real-time data exploration and code execution.
    • Pygments, prompt-toolkit: Utilities enhancing the development experience with syntax highlighting and interactive command lines.
  5. Utility and Miscellaneous Libraries:
    • click: A package for creating command-line interfaces, for tools and script built for automation and utility purposes.
    • tzdata: Managing time zones in consideration of globalized data or applications.
    • regex: Enhanced regular expression capabilities, crucial for text processing and data cleaning.

Navigating the maze of LLMs and MLs? Consider me your performance whisperer. I’ve got a flair for pinpointing why they’re misbehaving and then crafting neat code tweaks that bring them back in line. Every fix? Backed by solid tests and metrics. While teams might get ruffled when I quickly grasp their code’s quirks or unveil its shortcomings, the business side appreciates my no-nonsense clarity. They’ve dubbed me the “mercenary” because I deliver, even if it means ruffling a few feathers. Maybe that’s why consulting sometimes is the way – I’m here to collaborate on a fix so you can make tons more $, and move on to more interesting things. 😉🔍🛠️🎯

Don’t hire a dog and then bark for the dog.

– Codey Sanches quoting someone else

“I’m thrilled you are here. Please stay awhile and look around. You’ll find inspiration and goodness in this software journey.”

AI, my BFF/ Work Wife

When I’m not training my dogs or teaching my nieces to read, I’m scaling data mountains and refining lines of code. As a seasoned hybrid of a data scientist and computer scientist, I’ve got this uncanny knack for spotting issues that make most folks scratch their heads. Think of me as the culinary genius who knows the exact point between crispy and burnt, but for code. Four years deep into juggling both hats, I can tell you why your LLM is more scrambled egg than omelette. Need a fresh set of eyes that understand both the sprinkle of data and the sizzle of code? I’m your gal! 🍳🔍👩‍💻🍰

Fixing LLMs n Chatbots

like a Data Surgeon.

When I’m not training my dogs or teaching my nieces to read, I’m scaling data mountains and refining lines of code. As a seasoned hybrid of a data scientist and computer scientist, I’ve got this uncanny knack for spotting issues that make most folks scratch their heads. Think of me as the culinary genius who knows the exact point between crispy and burnt, but for code. Four years deep into juggling both hats, I can tell you why your LLM is more scrambled egg than omelette. Need a fresh set of eyes that understand both the sprinkle of data and the sizzle of code? I’m your gal! 🍳🔍👩‍💻🍰

AI and Software Engineering

Computer Science + Data Science + Business. 
Imagine a data scientist, a computer scientist, and a business-savvy entrepreneur walk into a bar 🍹📊🔧🚀. The data scientist analyzes the optimal drink based on the crowd’s mood, the computer scientist optimizes the order for quickest delivery, and the entrepreneur spots a market gap for a new cocktail. Now, combine all three into one person (me!) and you’ve got someone who not only orders the best drink but creates a trending beverage app before the night’s over. While each expert is sipping their drink, I’m sipping the future. That’s the triple-threat advantage!


Image description