Human-Centered AI

Decades of expertise across diverse domains including FinTech, HealthTech, InsurTech, Education, and Security.

Leveraging proven best practices from agile Silicon Valley startups to mission-critical Fortune 500 network security product development.

Integrating state-of-the-art AI research for data analytics, natural language processing, and multi-modal sentiment analysis.

That group (AI high-performers) has remained steady at about 8 percent… this group is achieving its superior results mainly from AI boosting top-line gains, as they’re more likely to report that AI is driving revenues rather than reducing costs, though they do report AI decreasing costs as well.

State of AI in 2022 — and a half decade in review, McKinsey (6 December 2022)

AI Large Language Models

LLM Research & Consulting

I have been researching LLM since GPT2 was first released in 2019, starting with fine-tuning models, generating diverse writings, and creating the first LLM Improv Divabot sponsored by the NEH.

I consult on state-of-the-art LLM applications in finance, healthcare, manufacturing, education and government. This includes benchmarking dozens of small 7B-14B models for privacy/security, using multiple prompt engineering strategies and controlling for limitations like hallucinations. The goal is application-driven end-to-end automated pipelines with clear ROI using LLMs, automated agents, tool chains and LangChain complemented with traditional ML/NLP techniques.

Co-Founded the first Human-Centered AI Curriculum

I co-founded the world’s first human-centered AI curriculum that is now a world center of excellence for human-centered AI research. We combine state-of-the-art AI research, best practices from industry, and a unique cross-disciplinary focus on real-word problems.

I have mentored over 300+ original research projects across virtually every discipline and medium including economics, social media, security, public policy, literature and many more. Within six years we have over 28k downloads from over 150 countries by prominent organizations like Stanford, MIT, Berkeley, CMU, Princeton, Oxford, Amazon, Intel, HP, Starbucks, Kate Spade, and over 2000 others wordwide.

NLP Sentiment Analysis

SentimentArcs & The Shapes of Stories

In 2021, SentimentArcs embarked on an ambitious project, aggregating and synthesizing an extensive suite of almost three dozen sentiment models. We introduced innovative metrics, data featurization, and normalization processes, all encapsulated within visually compelling representations, ensuring a streamlined human-in-the-loop supervisory workflow.

SentimentArcs serves as the cornerstone for the pioneering application of sentiment analysis in the realm of narratology, as showcased in ‘The Shapes of Stories,’ a seminal work by Dr. Katherine Elkins, published by Cambridge University. Using an ensemble of revered literary masterpieces, Dr. Elkins adeptly employs SentimentArcs to track the progression of emotions over time, extracting crucial insights from pivotal narrative junctures.