Members-Only
Recent Talks & Demos are for members only
You must be an AI Tinkerers active member to view these talks and demos.
May 26, 2026
·
Montreal
Claude-Powered Clinical Co-Pilot
See a clinical co-pilot using six Claude agents to detect drug interactions and bias. Learn how a trust boundary inside the process simplifies PHI handling and auditing.
Overview
A multi-agent clinical co-pilot that runs six specialist Claude agents in parallel on top of a PHI-safe pipeline : every LLM call sees only de-identified text, and re-identification happens server-side after the model returns.
Live, I’ll load a synthetic 68F new-AFib case where the cardiology plan proposes amiodarone for a patient already on warfarin. The Triage, Differential, Pharmacy, Guidelines, Bias-Check, and Communication agents stream into six panels via asyncio.gather. The Pharmacy agent flags the warfarin–amiodarone CYP2C9/3A4 interaction; Bias-Check independently flags anchoring on rhythm control; the orchestrator detects the cross-agent convergence and elevates it as a high-severity flag. I’ll then toggle the Safety Gate off so the audience sees raw PHI hit the LLM and get echoed back — then flip it on and show the audit log assertion (zero raw PHI, only counts + SHA-256 hashes). I’ll show the FastAPI/WebSocket code that fans out the agents, the trust-boundary diagram, the structured-field redactor, the live audit log tail, and the keystone pytest (test_pharmacy_catches_warfarin_amiodarone). Repo, logs, and 35/35 passing tests are all open.
Links
FastAPI-based clinical co-pilot utilizing regex, local summarizers, and Azure OpenAI.
Tech stack
- Claude CodeAnthropic's agentic coding tool: Unleash Claude's raw power directly in your terminal or IDE to turn complex, hours-long workflows into a single command.Claude Code is Anthropic’s powerful agentic coding assistant, designed for high-velocity development. It operates natively within your terminal, IDE (VS Code, JetBrains), or via a web interface, allowing you to delegate complex tasks like feature building, bug fixing, and codebase navigation. The agent plans, edits files, executes commands, and creates commits, maintaining awareness of your entire project structure. Internally, Anthropic engineers using Claude Code reported a 67% increase in productivity, demonstrating its capacity to deliver significant gains for Pro and Max plan users.
- FastAPIFastAPI is a modern, high-performance Python web framework for building APIs with automatic OpenAPI documentation.FastAPI is a robust, high-speed Python web framework: it is built on Starlette (for async capabilities) and Pydantic (for data validation and serialization). Leveraging standard Python 3.8+ type hints, the framework automatically generates interactive API documentation (Swagger UI/ReDoc) and enforces data validation, effectively reducing developer-induced errors by an estimated 40%. This architecture delivers performance on par with Node.js and Go, significantly increasing feature development speed (up to 300% faster). It is production-ready, fully supporting OpenAPI and JSON Schema standards for all API specifications.
- WebSocketWebSocket is the full-duplex communication protocol: it establishes a persistent, low-latency, bidirectional channel over a single TCP connection.WebSocket fundamentally changes client-server interaction from HTTP's request/response model by delivering persistent, full-duplex data exchange. The connection initiates via an HTTP/1.1 handshake (using the `Upgrade` header), then switches to the dedicated, stateful WebSocket protocol. This single, long-lived TCP connection (often on Port 443 for secure `wss://`) minimizes per-message overhead, eliminating the need for constant polling. The result is a low-latency, real-time data flow: essential for applications like live chat, multiplayer gaming, and financial trading dashboards.
- PythonPython: The high-level, general-purpose language built for readability, powering everything from web backends to advanced machine learning models.Python is the high-level, general-purpose language prioritizing clear, readable syntax (via significant indentation), ensuring rapid development for any team . Its ecosystem is massive: use it for robust web development with frameworks like Django and Flask, or leverage its power in data science with libraries such as Pandas and NumPy . The Python Package Index (PyPI) provides thousands of community-contributed modules, offering immediate solutions for tasks from network programming to GUI creation . The language is actively maintained by the Python Software Foundation (PSF), with the stable release currently at Python 3.14.0 (as of November 2025) .
- AzureAzure is Microsoft's global cloud platform: it delivers over 200 products and services (compute, networking, storage, AI) across a massive network of 60+ regions.Azure is the comprehensive cloud solution for modern business: we provide 200+ products, from IaaS (Virtual Machines, networking) to PaaS (Azure App Service, Azure SQL Database) and cutting-edge AI (Azure OpenAI Service). Our infrastructure spans more than 60 global regions and 140 availability zones, ensuring low-latency and high-availability for mission-critical workloads. You can deploy hybrid solutions with Azure Arc, manage enterprise identity with Microsoft Entra ID (formerly Azure AD), and scale applications on demand, all while maintaining compliance with over 100 certifications (e.g., ISO 27001, HIPAA). It’s the platform designed for secure, globally-scaled innovation.
- Azure OpenAIDeploy OpenAI's leading models (GPT-4, GPT-4o) with Azure's enterprise-grade security, compliance, and scalable infrastructure.Azure OpenAI Service delivers cutting-edge generative AI: models like GPT-4, GPT-4o, and DALL-E are accessible via secure REST APIs. This is not just a model wrapper; it integrates OpenAI's powerful capabilities with Azure's core enterprise features (data privacy, regional flexibility, Microsoft Entra ID authentication). Organizations confidently scale AI solutions (e.g., customer support automation, code generation) knowing their data remains private and is not used for model training. You get the best of both worlds: advanced AI and a trusted cloud foundation.
- GPT-4oGPT-4o (omni) is OpenAI's flagship multimodal model: it delivers GPT-4 intelligence with native, real-time processing across text, audio, and vision.This is GPT-4o, OpenAI’s 'omni' model: a single neural network natively handling text, audio, and image inputs and outputs. It matches GPT-4 performance on English text and code, but surpasses it on non-English language, vision, and audio benchmarks. The speed is a major upgrade: it achieves human-level responsiveness in voice, with an average response time of 0.32 seconds (a significant jump from GPT-4’s 5.4 seconds). Developers get a 128K token context window and a model that is more cost-efficient than its predecessor, making high-intelligence, real-time applications viable.
- AGENTSAutonomous software entities using large language models to reason, select tools, and execute complex workflows independently.Agents shift the focus from conversation to execution: they use frameworks like LangGraph or CrewAI to break down complex objectives into actionable tasks. These systems leverage external tools (Tavily for search, GitHub for code, or Salesforce for CRM) to operate across digital environments. Current benchmarks show agents can automate up to 80% of routine knowledge work by managing their own reasoning loops. These entities deliver finished outputs (validated data, resolved tickets, or deployed software) with minimal human intervention.
- StreamingReal-time delivery of digital media (audio, video) from a server to a client, enabling instant playback without full file download.Streaming is the continuous, sequential transmission of data packets, allowing a client device to play media immediately: no full file download required. This efficiency is powered by protocols like HLS (HTTP Live Streaming) and MPEG-DASH, which break content into small, indexed chunks. Global Content Delivery Networks (CDNs) are critical; they cache and distribute content across 300+ data centers, drastically reducing latency for services like Netflix, Spotify, and Twitch. Adaptive bitrate streaming is key to quality, dynamically adjusting the video resolution (e.g., from 480p to 4K) based on the user's current network speed to minimize buffering.
- AnthropicAnthropic is a frontier AI safety and research company, developing the Claude family of large language models (LLMs) via its Constitutional AI framework.Anthropic is an AI safety and research company, founded in 2021 by former OpenAI executives Dario and Daniela Amodei, and structured as a Public Benefit Corporation (PBC) . The core mission is building reliable, steerable AI systems, with a focus on interpretability and long-term alignment . Its flagship product is the Claude family of LLMs, which are highly capable models designed for complex reasoning and coding tasks . A key technical innovation is Constitutional AI (CAI), a training method that aligns the models with a set of ethical principles to ensure helpful, harmless, and honest outputs . The company has secured significant backing, including up to $4 billion from Amazon and a $2 billion commitment from Google .
- PromptPanic's premier SSH client for iOS and iPadOS delivers fast, secure terminal access from any mobile device.Prompt 3 provides a robust SSH environment for mobile power users. It features a custom terminal engine (supporting VT100 and Xterm-256color) and handles modern security protocols like Ed25519 and ECDSA keys. The interface includes a customizable accessory bar for essential keys (Esc, Tab, and arrows) to speed up command entry. Panic Sync keeps server lists and credentials consistent across iPhone, iPad, and Mac. Biometric locks (FaceID or TouchID) ensure keys stay protected while remaining accessible for rapid deployments.