Skip to content
View roydonsequeira's full-sized avatar
💭
Working from home
💭
Working from home

Organizations

@career-ops-hq

Block or report roydonsequeira

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.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. 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
roydonsequeira/README.md
Roydon Sequeira, AI Agent Developer at Code Crew Studio. Previously GenAI Engineer at ReinHealth, a healthcare AI startup. Based in Udupi, India. Open to AI engineering roles and contracts.

Portfolio: roydonsequeira.com LinkedIn X: @roydonsequeiraa Email

I build AI agents at Code Crew Studio, currently a WhatsApp onboarding agent on FastAPI and PostgreSQL. Before that I led a patient-intake agent at ReinHealth, a healthcare AI startup. It ran on local models, so no patient records went to outside APIs.

My open-source project is CORTEX, an agent that runs entirely on your own machine. Running it through 187 live conversations with a 7B model turned up 27 bugs its unit tests had missed. I wrote about what they taught me in 187 live prompts, 27 bugs.

Impact: about 70% less manual patient-intake work (estimated); patient intake down from 15 minutes to under 5; zero patient records sent to external APIs; 27 bugs in CORTEX found by 187 live conversations that its unit tests had missed.

Intake figures are estimates from the production agent I led at ReinHealth. The bug count comes from CORTEX's live test battery.

Open source

Most of these run entirely on your own hardware, with no API keys.

CORTEX, Private Intelligence Framework: a private AI agent that runs entirely on your own machine, with planning, sandboxed tools, four-tier memory, LATS tree search, a streaming UI and OpenTelemetry tracing, powered by Ollama.

CI status Last commit License: MIT

Why it holds up on a 7B model. Small models don't reliably follow rules that exist only in the prompt. In live testing with qwen2.5:7b, the model claimed it had saved a file without calling the tool, and followed an instruction hidden in text it was asked to summarize. So CORTEX enforces the rules in code:

  • Plans that need no tools run with no tool schemas at all, and destructive plans become refusals before anything executes.
  • Pasted or quoted text is treated as data, so instructions hidden inside it can't trigger a tool.
  • Web fetch reaches public hosts only. Loopback, private-network and cloud-metadata addresses are refused after DNS resolution and on every redirect.
  • Memory stores only durable facts you state about yourself.
Architecture and quick start
flowchart LR
    UI["Next.js UI"] -->|SSE| API["FastAPI gateway"]
    API --> K["Agent kernel<br/>plan · act · reflect"]
    K -. hard tasks .-> LATS["LATS tree search"]
    K -. big jobs .-> SUP["Supervisor → workers"]
    K --> MEM["Memory<br/>working · episodic · semantic · procedural"]
    K --> TOOLS["Sandboxed tools<br/>python · files · web · docs"]
    K --> R["Model router"] --> O["Ollama"]
    API -. OTLP spans .-> J["Jaeger"]
Loading
git clone https://github.com/roydonsequeira/CORTEX-Private-Intelligence-Framework.git && cd CORTEX-Private-Intelligence-Framework
ollama pull qwen2.5:7b && ollama pull nomic-embed-text
pip install -e . && cortex doctor && cortex serve    # API on :8000
cd ui && npm install && npm run dev                   # UI on :3000

RagChatbot: chat with your own PDFs, scans and Word files entirely on your machine, with OCR ingestion, local embeddings and streamed answers. Clinical SOAP Notes: n8n workflows that turn clinical text or voice into structured SOAP notes with a local LLM. Skin Lesion Segmentation: U-Net in PyTorch and ResU-Net in TensorFlow 2 on 2,594 ISIC 2018 dermoscopy images, with a Streamlit demo. In the lab, private builds: NEXUS AI content intelligence SaaS; a multi-agent code review agent; an AI resume analyzer with explainable ATS scoring and 80 tests; an AI proxy agent for WhatsApp, calendar and leads; a research agent with search, retrieval and citations.

Recently shipped

Project What it is Last push
CORTEX-Private-Intelligence-Framework
Python · ★ 6
Private, local-first AI agent: planning, sandboxed tools, four-tier memory, streaming UI and OpenTelemetry. Runs entirely on your machine with Ollama. 2026-11-17
Skin-Lesion-Segmentation-in-TensorFlow-2.0
Python
Skin lesion segmentation on ISIC 2018 using U-Net (PyTorch) and ResU-Net (TensorFlow 2). Final year project. 2026-10-18
RagChatbot
Python
Local RAG chatbot: Ollama LLM, ChromaDB vector store, OCR/PDF ingestion, Next.js UI 2026-04-09
clinicalNote-SOAP n8n workflows that turn clinical text or voice into structured SOAP notes with a local LLM 2026-03-25

Refreshed daily from the GitHub API by readme-sync.

Experience

  • AI Agent Developer, Code Crew Studio · Feb 2026 – present · remote
    Production onboarding agent on the WhatsApp Business API, with FastAPI and PostgreSQL for agent state and conversation history.
  • GenAI Engineer, ReinHealth (stealth healthcare AI startup) · Jul 2024 – Feb 2026 · remote, US
    Led a medical intake agent over text and voice on local Ollama models, with RAG on Qdrant and PostgreSQL, and n8n automations for clinical notes and scheduling.
  • Machine Learning Intern, Igeeks Technologies · Jun – Jul 2023 · Bengaluru
    CNN, AlexNet and MLP image classifiers with OpenCV preprocessing.

Education: B.E. in Artificial Intelligence & Machine Learning, NMAM Institute of Technology (2020–2024) · Executive PG Certification in Data Science & AI, iHUB DivyaSampark, IIT Roorkee (2024–2026)

Stack

Layer Tools
Agents & LLMs LangChain · ReAct · Ollama · Claude · Gemini · OpenAI · Hugging Face · STT/TTS
Retrieval Qdrant · ChromaDB · BGE and nomic embeddings · semantic search · PostgreSQL · Redis
Backend Python · FastAPI · Flask · Pydantic · SQL · REST · SSE streaming
Automation n8n · WhatsApp Business API · Twilio
ML & vision PyTorch · TensorFlow · scikit-learn · OpenCV · NumPy · Pandas
Ship & observe Docker · Linux · GitHub Actions · OpenTelemetry · Vercel · ruff · mypy
Agent UIs Next.js · TypeScript · Tailwind CSS

Open to AI engineering roles and contract work: roydnsequeira@gmail.com

Popular repositories Loading

  1. CORTEX-Private-Intelligence-Framework CORTEX-Private-Intelligence-Framework Public

    Private, local-first AI agent: planning, sandboxed tools, four-tier memory, streaming UI and OpenTelemetry. Runs entirely on your machine with Ollama.

    Python 6 2

  2. roydonsequeira roydonsequeira Public

    Python

  3. Skin-Lesion-Segmentation-in-TensorFlow-2.0 Skin-Lesion-Segmentation-in-TensorFlow-2.0 Public

    Skin lesion segmentation on ISIC 2018 using U-Net (PyTorch) and ResU-Net (TensorFlow 2). Final year project.

    Python

  4. clinicalNote-SOAP clinicalNote-SOAP Public

    n8n workflows that turn clinical text or voice into structured SOAP notes with a local LLM

  5. RagChatbot RagChatbot Public

    Local RAG chatbot: Ollama LLM, ChromaDB vector store, OCR/PDF ingestion, Next.js UI

    Python

  6. ollama ollama Public

    Forked from ollama/ollama

    Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.

    Go