[ Portfolio ]

Projects

Explore_by_Focus_Area
01 /  Clinical & Biomedical Data Science

Transcriptomic & Immune Microenvironment Profiling of Bladder Cancer Progression

OPPORTUNITYThe biology separating a contained bladder tumor from one that invades is unresolved, yet that single distinction drives entirely different treatment paths.

WHAT I BUILTA four-stage R pipeline — differential expression, functional enrichment, immune deconvolution, and longitudinal visualization — tracing molecular and immune change from pre-cancer to muscle-invasive disease.

IMPACTSurfaced an EMT/ECM remodeling signature at tumor onset and a TNF-driven inflammatory shift at invasion, plus immune-composition changes relevant to immunotherapy timing and patient selection.

RBIOINFORMATICSMICROARRAY
Mendel logo — Your health, decoded
02 / PRODUCTProduct & Platform Development

Mendel — Your Health, Decoded

OPPORTUNITYConsumers receive dense genetic and health data they can't interpret or act on.

WHAT WE BUILTA consumer health product concept that translates genomic results into clear, prioritized, actionable guidance — built as a team.

IMPACTEnd-to-end product and UX design connecting complex genomics to everyday health decisions.

PRODUCTUX DESIGNGENOMICS
03 / BIOINFORMATICSClinical & Biomedical Data Science

Gene Expression Biomarkers Across Alzheimer's Disease Progression

OPPORTUNITYAlzheimer's affects millions but has no confirmed cause or disease-modifying treatment; gene expression from post-mortem brain tissue could reveal biomarkers for staging or targets.

WHAT WE BUILTA differential expression and multi-database pathway enrichment pipeline in R, comparing moderate vs. incipient and severe vs. moderate AD stages, with literature-grounded interpretation of four candidate pathways.

IMPACTIdentified four biologically coherent themes — estrogen signaling loss, insulin resistance, neuroinflammation, and mitochondrial dysfunction — that track with disease severity and are supported by primary literature.

RBIOINFORMATICSPATHWAY ENRICHMENT
Health data discovery platform dashboard with dataset quality scoring
04 / PLATFORMProduct & Platform Development

Real-World Psychiatric Data Commons

OPPORTUNITYReal-world evidence lives in fragmented, poorly documented datasets, making discovery slow and quality hard to trust.

WHAT I BUILTA full-stack platform for ICA.ai that unifies RWE datasets behind AI-powered discovery, automated FAIR quality scoring, and real-time team collaboration.

IMPACTCut data discovery time by 75% while making dataset quality visible and comparable across teams.

FULL-STACKFAIRAI SEARCH
05 / NLPClinical & Biomedical Data Science

Classifying Smoking Status from Unstructured Clinical Notes

OPPORTUNITYSmoking status is buried in free-text discharge notes, making it costly to extract for research cohort building, quality reporting, and risk stratification.

WHAT I BUILTA complete clinical NLP pipeline from raw XML to a fine-tuned transformer, combining bag-of-words features, domain-specific keyword injection, ADASYN rebalancing, grid-searched classical ML, and ClinicalBERT fine-tuning.

IMPACTA methodologically sound end-to-end pipeline that correctly diagnoses severe class imbalance as the central failure mode rather than inflating accuracy.

NLPCLINICALBERTPYTHON
06 / HEALTHCARE QIClinical & Biomedical Data Science

Streamlining STEMI Care: A Database-Driven Quality Improvement Project in a Synthetic EHR

OPPORTUNITYDoor-to-balloon delays in STEMI care directly worsen mortality, yet the workflow was reactive and manual — a classic emergency cardiology quality problem.

WHAT I BUILTA full QI cycle: SQL cohort construction from a synthetic EHR, three live OpenEMR clinical decision support rules, and a multi-angle evaluation covering cost, workflow timing, and staff sentiment.

IMPACTDemonstrated a functioning alert-driven intervention with honest adoption findings (ease-of-learning gap) that directly shaped an incremental rollout recommendation.

SQLOPENEMRHEALTHCARE QI
07 / MACHINE LEARNINGWearable Health & Fitness

Wearable Sensor Stress Detection w/ Spotify

OPPORTUNITYStress is continuous and physiological, but most interventions are manual and arrive far too late.

WHAT I BUILTAn ML pipeline that classifies stress from wearable sensor signals and integrates with Spotify to trigger calming, adaptive playlists.

IMPACTA closed-loop wellness prototype that senses stress and responds in the moment with music therapy.

MACHINE LEARNINGWEARABLESSPOTIFY API
Smartwatch stress detection interface showing elevated stress score with adaptive music playlist
Side_Work

On Going Work

Continued work that combines health domain knowledge with product thinking 

S1

VIDA Fitness App Concept

A self-directed product concept for the fitness and wellness space, focused on product thinking and translating health domain knowledge into a clear user-facing experience.

PRODUCTUXWELLNESS
S2

Wearable Data Insight Layer

A visualization tool concept that turns raw wearable device data into a clear, actionable “so what” layer for the end user, rather than just displaying raw metrics.

DATA VIZWEARABLESINSIGHT
S3

Personal Financial Data Pipeline

An end-to-end pipeline connecting a financial data provider, a local database, and a custom integration layer to build automated, structured insight from raw transaction data.

PIPELINEETLAUTOMATION
S4

Applied AI Development Workflow Experiments

Ongoing experimentation with AI-assisted development workflows — pairing AI coding tools in a build-and-review loop to explore how AI can improve software speed and quality.

AI TOOLINGDXWORKFLOW