Selected work

Projects and portfolio themes

Selected work themes across healthcare, finance, and enterprise teams, centered on production AI systems, evaluation, data products, and experiment infrastructure that move beyond demos.

Explore AI Engineering Read LLM Evaluation Data Products Discuss Collaboration
Project grid

Featured project and case-study areas

These cards organize the site around concrete applied AI and data product work. They preserve crawlable internal links to the deeper technical notes.

LLM Systems

Applied AI for document and knowledge workflows

Problem: teams need AI systems that can use documents, retrieve evidence, and support knowledge work without becoming opaque answer machines.

Methods and tools

Retrieval-augmented generation, document intelligence, report generation, MCP-style integration, prompt and schema design.

Outcome or value

Better route from source material to reviewable AI outputs that can be integrated into real team workflows.

AI Engineering column LlamaReport guide MCP guide
AI Evaluation

Evaluation systems for LLM behavior and uncertainty

Problem: model outputs need task-specific evaluation, uncertainty awareness, and regression checks before they can support high-stakes decisions.

Methods and tools

LLM evaluation design, uncertainty estimation, rubric-based checks, adversarial examples, human review workflows.

Outcome or value

Clearer evidence about when an AI system is reliable, when it should defer, and what should be improved before launch.

LLM Evaluation Testing Copilot Agents UQLM teaching guide
Healthcare Analytics

Decision-support analytics for healthcare and operations

Problem: operational teams need analytical systems that translate messy data into decisions, not just charts or one-off notebooks.

Methods and tools

Statistical analysis, data products, dashboard design, cohort analysis, measurement planning, quality controls.

Outcome or value

More usable analytics workflows for clinical, operational, or enterprise teams that need trustworthy decision support.

Healthcare AI Analytics Statistical tests guide Data Products
Data Products

Products that move teams from raw data to decisions

Problem: data work often fails when it stops at analysis and never becomes a product teams can repeatedly use.

Methods and tools

Data product strategy, analytics UX, product requirements, experimentation, data engineering, decision-support design.

Outcome or value

Reusable systems that connect data, product workflows, and operational decisions with clearer adoption paths.

Data Products Product strategy Product fundamentals
NLP / Topic Modeling

Machine learning and NLP for interpretable text systems

Problem: text-heavy workflows need models that classify, summarize, cluster, and explain language patterns in ways humans can inspect.

Methods and tools

NLP, topic modeling, sentiment analysis, statistical ML, trustworthy machine learning, model interpretation.

Outcome or value

More explainable text analysis for research, product, and enterprise workflows where patterns need to be understood.

Machine Learning & NLP BERT sentiment analysis Trustworthy ML
Experiment Infrastructure

From experimentation to production delivery

Problem: teams need reliable ways to test ideas, compare outcomes, and monitor launches without losing engineering discipline.

Methods and tools

A/B testing systems, MLOps, data platforms, launch measurement, CI/CD thinking, observability, delivery systems.

Outcome or value

Better path from experiment to production, with measurement and infrastructure that make iteration safer and clearer.

A/B testing systems MLOps foundations Data platform guide

For a broader professional overview, see the resume overview; for public writing by topic, use the blog index.