Dan Landayan

Scientist

Neuroscientist → ML Research Engineer

Connectomics‑driven machine learning in PyTorch. I build research‑to‑production pipelines that turn large wiring diagrams into models, insights, and demos.

San Leandro, CA – open to Bay Area roles

Data & ML

PyTorch, scikit‑learn, GNNs & RL

SQL, pandas & cloud warehouses

Hugging Face, transformers

Domain

Connectomics & neurobiology

Genomics & single‑cell omics

Graph & gene network priors

Engineering

Reproducible pipelines & CI/CD

Docker, AWS SageMaker & FastAPI

Distributed training & monitoring

Communication

Research papers & blog posts

Clear READMEs & storytelling【10811468707479†L14-L15】

Teaching & open source contributions

Featured Projects

FlyWire Connectomics Toolkit

fafbseg‑py + navis for programmatic connectome analysis
fafbseg-pynavisPyTorchDockerCI
  • 1,000+ neurons analyzed

Programmatic access to FlyWire via fafbseg‑py, standardized coordinate transforms, navis interoperability, and a small PyTorch module for learned morphology/connectivity embeddings.

  • Neuron/neuropil retrieval + transforms; reproducible notebooks → Dockerized pipeline
  • navis‑compatible objects for downstream morphology/graph analyses
  • Unit tests & CI surfaced in repo; paper‑style README

oviIN Input‑Module Analysis

Community detection on oviIN subconnectome + PyTorch embeddings
Graph MLPyTorchStreamlitDockerCI
  • 1,000+ presynaptic partners

Reproduces the oviIN multi‑circuit hub input analysis: modular structure over presynaptic partners, PyTorch embeddings for similarity metrics, and a Streamlit demo to browse modules.

  • Modularity on oviIN inputs; per‑module statistics + synapse maps
  • Abstract→Methods→Results→Discussion; figures auto‑rendered
  • Streamlit exploratory UI; Docker env; CI badge

Research → Production

I believe high‑impact science comes from pairing curiosity‑driven research with rigorous engineering. Each project follows a structured lifecycle that reflects this philosophy【0†L20-L25】:

Reproducible Pipelines

Clean repos with dependency management, version control and documentation【0†L3-L8】.
  • Environment files (conda, pip)
  • Clear data & code separation (src/, data/, notebooks/)
  • Automated data preprocessing scripts

Testing & CI/CD

Light unit tests and continuous integration show engineering maturity【0†L3-L8】.
  • pytest suites cover core functions
  • GitHub Actions for linting & tests
  • Pre‑commit hooks & code quality tools

Deployment & Monitoring

Bringing models into production with proper observability【0†L3-L8】.
  • Docker images & container registries
  • FastAPI or SageMaker endpoints
  • Logging, metrics & cost optimization

Publications & Teaching

Beyond code, I share knowledge through papers and lectures. Recent highlights include:

  • Fruit Fly Decision‑Making: First author on a study exploring how multiple circuits converge in oviIN neurons to govern state‑dependent egg‑laying behaviour – inspiring modular architectures in RL【0†L18-L25】.
  • Adult Fly Connectome Teaching: Developed course material and hands‑on labs for mapping the Drosophila connectome using navis/fafbseg‑py and demonstrating graph ML applications.
  • Guest Lectures: Hosted workshops on integrating multi‑omics data into ML models and wrote blog posts on neuro‑inspired architectures【224711995524461†L288-L299】.

Let’s collaborate

Whether you’re building the next generation of biomedical AI products or exploring fundamental questions at the intersection of computation and biology, I’d love to help.

Fruit Fly Foraging Connectomics

Reinforcement-learning environment built on FlyWire connectome data.
  • Leverages full Drosophila brain connectome (FlyWire) to simulate realistic neural inputs.
  • Central complex circuits modeled to drive a virtual fly’s foraging decisions.
  • Uses RL agents to test circuit hypotheses and generate behavioral predictions.