Skip to content
View DimiHepburn's full-sized avatar

Highlights

  • Pro

Block or report DimiHepburn

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 is supported. This note will only be visible to you.
Report abuse

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

Report abuse
DimiHepburn/README.md
Typing SVG

Profile Views


🧠 About Me

I'm a researcher and builder working at the intersection of neuroscience and artificial intelligence. Currently studying MSc Clinical Neuroscience at the University of Roehampton (London), with a background in Clinical Psychology (BSc Hons, London South Bank University).

My work spans from building real AI systems β€” autonomous research agents, custom hardware assistants, and MCP-integrated workflows β€” to academic research in neuroinflammation, the gut-brain axis, and psychopharmacology.

  • πŸ”§ Built a custom AI assistant on ESP32-S3 hardware with self-hosted inference and Anthropic API integration
  • πŸ€– Developed an autonomous research agent using a ReAct reasoning loop with PubMed search and importance-weighted memory
  • πŸ“ Authored a tetrad of frameworks for human-centred AI β€” Friction Protocol, Humanising Loop, Attunement Audit, and Handoff Threshold
  • 🧬 Designed the MICROBE-D Study β€” a proposed RCT investigating Bifidobacterium breve CCFM1025 for depression
  • πŸ’¬ Ask me about: neuro-AI, psychopharmacology, LLM interpretability, brain-computer interfaces, gut-brain axis

πŸ”­ Research Interests

  • 🧠 Clinical Neuroscience β†’ Neuroinflammation, BBB dysfunction, glymphatic system
  • πŸ’Š Psychopharmacology β†’ Drug-brain interaction modelling, MAOI mechanisms
  • 🦠 Gut-Brain Axis β†’ Microbiome-mediated mechanisms in neuropsychiatric conditions
  • 🀝 AI Humanisation β†’ Affective computing, empathetic AI, conversational agents
  • πŸ” Mechanistic Interpretability β†’ Understanding what happens inside transformers
  • ⚑ Brain-Computer Interfaces β†’ EEG-based systems, neural signal processing

πŸ› οΈ Tech & Tools

Python PyTorch Anthropic ESP32 LiteLLM MCP NumPy Pandas SciPy Jupyter Git macOS


πŸ“Œ Featured Projects

Project Description Status
πŸ€– Sol AI Assistant Custom AI assistant on DeepSeek XiaoZhi ESP32-S3 hardware, self-hosted via xiaozhi-esp32-server with LiteLLM β†’ Anthropic API routing βœ… Built
πŸ”¬ ReAct Research Agent Autonomous Python agent with ReAct reasoning loop, Anthropic API, PubMed search integration, and importance-weighted memory store βœ… Built
πŸ“ Friction Protocol Original pedagogical framework for AI-augmented learning β€” bridging Polanyi, SchΓΆn, Floridi, and Vygotsky πŸ“„ Published in coursework
🦠 MICROBE-D Study Proposed 12-week RCT: B. breve CCFM1025 as adjunctive treatment for mild-to-moderate depression, featuring 16S rRNA sequencing and inflammatory biomarkers πŸ“„ Study design complete
🧠 neuro-ai-bridge Mapping neuroscientific principles onto deep learning architectures β€” Hebbian learning, SNNs, predictive processing, hippocampal memory πŸ”§ Active research
🀝 humanising-ai A tetrad of frameworks for emotionally intelligent, human-centred AI β€” Friction Protocol, Humanising Loop, Attunement Audit, Handoff Threshold β€” with reference implementation across affective, theory-of-mind, dialogue, and explainability modules πŸ”§ Active research
πŸ” llm-interpretability-notes Research notes and experiments on mechanistic interpretability of LLMs β€” residual stream, attention circuits, SAEs, activation patching πŸ”§ In progress
πŸ“– neuro-readability Neuroscience-informed CLI tool scoring text for cognitive load and readability 🐍 Python

🧭 Research Programme

The three research repositories form a deliberate triangle:

  • neuro-ai-bridge β€” what brains do β†’ biological learning mechanisms as inductive bias for AI
  • llm-interpretability-notes β€” what models do β†’ reverse-engineering the internal computations of transformers
  • humanising-ai β€” what models should do β†’ frameworks and implementations for AI that behaves in a way that is legible, attuned, and safely bounded

Together they triangulate a single question: how do we build AI systems that are both mechanistically understood and humanely aligned?


πŸ“Š GitHub Stats

Dimi's GitHub stats Top Languages


🌱 Currently Exploring

  • Brain-computer interfaces and EEG signal processing for real-time neurofeedback
  • Spiking Neural Networks (SNNs) as biologically plausible learning systems
  • The role of attention mechanisms as analogues to human selective attention
  • Ethical frameworks for emotionally responsive AI
  • Transformer interpretability via activation patching and probing classifiers
  • MCP server integrations for research workflow automation

πŸŽ“ Academic Background

MSc Clinical Neuroscience β€” University of Roehampton (2025–present) Targeting Distinction | Focus: neuroinflammation, psychoneuroimmunology, psychopharmacology Key outputs: BBB & glymphatic dysfunction in Huntington's disease, B cells in MS, microglia in MND

BSc (Hons) Clinical Psychology β€” London South Bank University (Graduated 2025) Dissertation on empathy and emotional contagion in couples (2.2)


πŸ“« Get in Touch

I'm open to research collaborations, PhD opportunities in neuro-AI or psychopharmacology, and research assistant roles. Always happy to nerd out about consciousness, computation, and the spaces in between.


"The brain is wider than the sky." β€” Emily Dickinson

Popular repositories Loading

  1. testflightybroadcastgroups testflightybroadcastgroups Public

    1

  2. testflightybroadcastgroups2 testflightybroadcastgroups2 Public

    1

  3. DimiHepburn DimiHepburn Public

    Profile README β€” MSc Clinical Neuroscience researcher working at the intersection of neuroscience and AI

  4. neuro-ai-bridge neuro-ai-bridge Public

    Mapping neuroscientific principles onto modern deep learning architectures β€” from Hebbian learning to spiking neural networks

    Python

  5. humanising-ai humanising-ai Public

    Research and frameworks for building emotionally intelligent, empathetic, and genuinely human-centred AI systems

    Python

  6. llm-interpretability-notes llm-interpretability-notes Public

    Research notes, experiments and paper summaries on mechanistic interpretability of large language models

    Python