Clinical AI & NLP
Real-world EHR research using MIMIC-III, clinical risk modeling, medical NLP, evaluation, and reproducible cohort construction.
MIMIC-IIIBigQueryXGBoostClinical NLP
Research · Engineering · Evidence
Building AI systems that can be tested, trusted, and used in the real world.
I am an Applied AI Research Engineer focused on healthcare AI, trustworthy AI, and real-time intelligent systems. My work combines rigorous experimentation, reproducible engineering, model governance, and practical deployment thinking.
One research profile. Multiple real-world domains.
I work at the intersection of machine learning, software systems, and scientific validation. The same discipline that makes AI safer in healthcare also matters in finance, cybersecurity, enterprise AI, and other high-consequence environments.
AI Evaluation & Governance
Leakage controls, model reproducibility, prospective evidence, fail-closed decision gates, negative-result reporting, and human governance.
ValidationGovernanceReproducibilitySafety
Real-time Intelligent Systems
Streaming market data, distributed services, online scoring, telemetry, event pipelines, risk controls, and evidence capture under latency constraints.
PythonDockerCloudStreaming
Selected research work
These projects are presented as research evidence—not marketing claims. Each case study separates the question, data, methodology, validation, results, limitations, and next experiment.
Flagship 01 · Healthcare AI
Critical-Care Risk Prediction with Real MIMIC-III Data
A reproducible sepsis-focused research workflow using 14,966 adult patients, first-24-hour ICU features, supervised learning, and evaluation of mortality prediction. The work uses real PhysioNet MIMIC-III data through BigQuery and includes structured clinical variables plus separate clinical NLP studies.
14,966 patients15 clinical featuresLogistic RegressionRandom ForestXGBoost
Flagship 02 · Autonomous AI Systems
Alpha Yazan AI — Governance-First Strategy Research
An experimental real-time research platform developed within Yazan Alpha Technologies, Inc. for generating, testing, falsifying, and governing candidate market strategies. The emphasis is on evidence collection, reproducibility, risk authority, shadow evaluation, fail-closed safety, and explicit promotion controls.
Real-time dataModel governanceShadow researchSafety controlsEvidence pipelines
Flagship 03 · Research Direction
Trustworthy Clinical AI Research Platform
A planned research direction for multimodal clinical prediction and safe clinical NLP: structured EHR signals, notes, temporal features, calibration, uncertainty, fairness, and explicit evidence boundaries before any deployment claim.
Clinical deteriorationRAG evaluationCalibrationFairness
Research operating model
Problem → hypothesis → experiment → evidence → decision
Real-world problem
→
Testable hypothesis
→
Experiment design
→
Evidence & validation
→
Failure analysis
→
Decision / next test
The goal is not to make every experiment succeed. The goal is to make every important conclusion traceable to evidence.
Built for collaboration
I am interested in fully remote applied AI research and research-engineering roles, healthcare AI collaborations, hospital/health-tech research partnerships, and evidence-driven AI projects where scientific rigor matters as much as model performance.