Health Data Scientist & Builder
I build systems that move cells, proteins, and clinical training from experiment to something another person can run.
PyTorch
OpenAI
Cursor
Flask
MongoDBPyTorchOpenAICursorFlaskMongoDBFocused on health data science, computational biology, and tools other people can run. I learn by shipping — a simulator, a CLI, a cluster loop.
A GPT-based rapid-cycle simulator for pediatric airway training, built with Emory Pediatrics and Children's Healthcare of Atlanta. Residents run cases. Admins edit instructions.
Predict human cells from mouse cells with optimal transport in a shared autoencoder latent space. One CLI lists models, prints a scorecard, and shows the sbatch chain a human still submits.
$ ./hub list atlas_full / impact_cellot atlas_full / scgen $ ./hub scorecard frac_gap_closed_decoded guardrails: means · JS $ ./hub show <run_id> sbatch chain ready
Fairshare-aware ablation search on FASRC Cannon: submit, watch, reflect, decide. A brain proposes the next agenda. A substrate keeps executing if the laptop sleeps.
Health Data Stack
A focused map of the tools behind the simulator, the transport CLI, and the cluster loop — shaped by systems I actually build and run.
Python · PyTorch · scanpy · CellOT · AlphaFold · R
React · Flask · MongoDB · Auth0 · AWS
OpenAI · Cursor · LLM directors · autonomous search loops
SLURM · FASRC Cannon · fairshare · HPC
Write if the work is useful — a collaboration, a question about a repo, or a pointer to something I should see.
[email protected]