Skip to content

CodeHalwell/CodeHalwell

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 

Repository files navigation

👋 Hey, I’m Dan (CodeHalwell)

Analytical Chemist → Data Scientist / AI Engineer
Currently poking at large language models and automation at scale.
Based in the rainy half of the UK, trying to use AI for things that actually matter.


🧪 What I do

  • Build tooling around LLMs (config, orchestration, agents, RAG)
  • Bridge scientific thinking with data & AI in pharma and healthcare
  • Prototype ideas fast, then slowly make them less embarrassing
  • Care about responsible, human-centred, climate-conscious tech rather than “move fast and break society”

🔭 Current focus

  • yamllm – YAML-first configuration for LLMs so you can swap models, prompts and memory without rewriting half your code
  • Agentic AI – workflows, tools and MCP servers for actually useful agents rather than “yet another chat box”
  • RAG & evaluation – small, focused retrieval systems and ways to check if the model is talking sense
  • Developer ergonomics – scripts, CLIs and templates to make working with LLMs less fiddly and more reproducible

📌 Pinned work

Agent frameworks & orchestration

  • AgentGuides
    A big, opinionated set of guides for building AI agents across 16+ frameworks & SDKs
    (OpenAI Agents, SmolAgents, CrewAI, LangGraph, LlamaIndex, PydanticAI, Bedrock, Claude, etc.) – with deployment patterns, diagrams and copy-pasteable recipes.

  • yamllm
    A Python library for YAML-first LLM configuration & execution.
    Keep prompts, models, tools, memory, parameters and routing in config instead of burying them in scripts. Good for reproducible experiments and swapping providers without rewiring everything.

Research & assistants

  • deep-research
    An agent-based research workflow built on LlamaIndex and MCP.
    Multi-agent pipeline (planning → research → writing → review → formatting → summary) with web/academic search, PDF generation and human-in-the-loop quality gates.

  • gradio-mcp-agent-hack
    A multi-agent “shallow research” + code assistant wired through a Gradio MCP server.
    Integrates with VS Code / MCP clients, runs generated Python in sandboxed Modal environments, and focuses on grounded research plus executable answers.

Interfaces & personal tooling

  • code_chat_bot
    Multi-provider Streamlit LLM app (OpenAI, Mistral, Anthropic, Cohere).
    Chat about code, load documents/URLs, swap models mid-conversation, and keep an eye on token spend.

  • digital-cv
    An AI-powered digital CV built with Gradio.
    Lets people chat with “me” about experience, projects and skills; supports contact capture, question tracking and analytics, all driven from my actual CV and profile docs.


🛠️ Toolbox

Languages & data

  • Python, little bit of SQL, little bit of TypeScript, a dash of notebook-driven experimentation
  • Pandas, scikit-learn, Jupyter, PyTorch, TensorFlow, transformers

LLMs & agents

  • OpenAI, Anthropic, Mistral, etc. via APIs
  • LlamaIndex, LangGraph, various agent frameworks
  • Model Context Protocol (MCP) servers & clients
  • RAG systems, research agents, evaluation workflows

Apps & orchestration

  • Streamlit & Gradio for quick UIs
  • n8n & workflow-style automation
  • Docker, uv, virtualenvs and the usual Python plumbing

🧠 How I think

  • Science first: hypotheses, experiments, evidence
  • Data & models should be interpretable, reproducible and testable
  • Prefer open standards and open source over lock-in where possible
  • Tech should make people’s lives better, not just someone’s quarterly earnings

🌱 Outside the code

When I’m not coding, I’m usually:

  • Reading about AI, systems design, economics, science
  • Exploring how we can use AI to improve our lives
  • Refactoring yesterday’s “quick hack” into something future-me might forgive

📬 Find me

If you’re into LLM tooling, AI in science, or just making data work a bit harder for humans, feel free to reach out or open an issue/PR.

About

My profile page README file

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors