Skip to content
View NickLeko's full-sized avatar

Block or report NickLeko

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

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

Report abuse
NickLeko/README.md

Nicholas Leko

Healthcare AI builder focused on workflow automation, interpretable ML, and evaluation safety.

I build healthcare AI projects that are scoped, explainable, and honest about their limits. My portfolio is centered on three layers of the stack:

  • Workflow support
  • Interpretable prediction
  • LLM evaluation and safety

Start here

Safety-first administrative decision support for prior authorization readiness.

What it shows

  • Healthcare workflow realism
  • Deterministic scope boundaries
  • Requirement-level evidence mapping
  • Governance and policy-drift awareness

Interpretable ICU deterioration-risk modeling built as a reproducible scientific artifact.

What it shows

  • Transparent ML in a healthcare setting
  • Reproducibility and maintenance discipline
  • Honest evaluation and artifact governance
  • Controlled use of AI coding tools around sensitive ML logic

A safety-focused evaluation harness for clinical-style LLM outputs.

What it shows

  • Faithfulness and citation-aware evaluation
  • Uncertainty and refusal analysis
  • Failure interpretation over model hype
  • Benchmark discipline and clear non-claims

What ties these projects together

Across all three projects, the common thread is the same:

  • Narrow, defensible scope
  • Explainable system behavior
  • Strong documentation and reviewer clarity
  • Reproducibility and maintenance boundaries
  • Healthcare AI judgment over generic demos

Current focus

  • Healthcare AI product management
  • Admin workflow automation
  • LLM evaluation and safety
  • Interpretable ML in clinical settings
  • Building portfolio artifacts that are useful in real workflows, not just technically interesting

Contact

Pinned Loading

  1. PriorAuthorizationCopilot PriorAuthorizationCopilot Public

    Administrative decision-support system for prior authorization readiness. Deterministic, rules-first evaluation of documentation completeness with refusal semantics, full auditability, and write-on…

    Python

  2. clinical-AI-eval_sandbox clinical-AI-eval_sandbox Public

    A lightweight evaluation framework that simulates how a healthcare company might risk-test an LLM before deploying it into clinical decision-support workflows.

    Python

  3. icu-code-blue-early-warning icu-code-blue-early-warning Public

    Early warning ML pipeline for ICU cardiac arrest risk using eICU-CRD with structured feature engineering and model evaluation.

    Shell