// AI Product Leader · Group PM · Builder · Miami Beach, FL
Ralph Charles Evert IV
I'm Charlie Evert — an AI product leader and engineering builder who'd rather ship than pitch. Equally comfortable owning a roadmap, writing production code, leading a 50-person delivery team, or sitting in front of a C-suite to defend the tradeoffs. 6+ years spanning product management, engineering leadership, and AI strategy across healthcare, life sciences, federal, and Fortune 500. MBA + MS in Data Science.
What I actually do: identify high-value opportunities, translate complex workflows into scalable AI products, and deliver measurable business impact. End-to-end development of knowledge retrieval systems, document automation tools, multichannel agentic platforms, and decision-support systems — used by 1M+ users and driving $100M+ in value. Comfortable partnering with R&D, regulatory, scientific, and data science stakeholders to ship compliant production-grade GenAI.
My background as a developer, tech lead, group product manager, program manager, and strategy consultant lets me move fluidly between code, requirements, roadmaps, and stakeholder rooms — going from interview → real problem → tech fix → measurable business outcome. Always open to connect.
Also a published author, US Patent holder, peer-reviewed LLM-evaluation researcher (Anthropic × UVA), and AI-themed metal music producer because why not.
I came up writing production code and still ship it. Architect systems on RAG, knowledge graphs, agentic workflows, multi-channel orchestration, and LLM eval — and lead the engineers who build them.
Define and execute roadmaps for portfolios of GenAI products — currently 15+ at PwC. Previously Group PM for a 5-product Human Capital AI Suite at Deloitte ($300M+ pipeline).
Identify high-value opportunities, translate complex workflows into AI products, and defend technical tradeoffs to C-suite stakeholders without losing the engineering thread.
Directed 100+ engineers, designers, and PMs across 20+ concurrent AI projects in matrixed environments. Fluent translating tech tradeoffs to scientific, clinical, and operational stakeholders.
Airborne Infantryman, Fort Benning, GA. Operates under pressure with incomplete information. Has jumped out of planes.
Author of 3 published books, 5 peer-reviewed papers (incl. UVA × Deloitte LLM evaluation research with Anthropic), US Patent Pending, YouTube channel.
Most AI products fail not because the model was wrong, but because nobody framed the right problem, defined the right metrics, or made the tradeoffs explicit. I run product end-to-end — discovery through measurement — and I'm comfortable picking up code, slides, or stakeholder maps depending on what the moment needs.
Interview scientific, clinical, and operational stakeholders. Frame the actual problem — not the feature anyone asked for. Size the opportunity and surface unstated compliance, traceability, and risk constraints.
Translate workflows into roadmaps, user stories, requirements, ROI models, and adoption KPIs. Sequence programs against value cases. Choose agile, waterfall, or a hybrid based on regulatory posture and engineering reality.
Lead engineering hands-on as tech lead and group PM. Make architecture decisions on RAG, knowledge graphs, agentic workflows, and evaluation. Ship to production in regulated environments — not the whiteboard.
Track adoption, P&L impact, and quality against the ROI model from step 2. Iterate. Kill what isn't working. Communicate results to senior stakeholders so investment decisions stay aligned to outcomes.
A sampler of the GenAI products I've owned, scoped, shipped, or led engineering on across regulated R&D, federal health, and Fortune 500 enterprise. Where I was the PM, you'll see Owned roadmap; where I was the tech lead, Led engineering; where I led the team that built it, Engagement lead.
First enterprise GenAI platform at J&J Consumer Health — agentic text-to-SQL + Wolfram + RAG chatbot. Originated inside R&D and scaled enterprise-wide. Shipped 2 months after ChatGPT launch.
Scientific knowledge platform for OTC R&D scientists: packaging compatibility checks, regulatory warnings, predictive chemical-attribute models, and competitor-formula similarity scoring. Data scraped from PubChem & CAS.org; competitor cost-mapping vs. Amazon prices.
Enterprise-wide GenAI assistant for performance management. Featured in Authority Magazine as a flagship enterprise AI win.
Agentic document-generation system replacing manual report compilation for federal regulatory submissions. Navigated FedRAMP, traceability, and audit-readiness requirements characteristic of regulated drug-development workflows.
Large-scale AI knowledge platform for a $1B+ nonprofit, translating regulatory and user needs into a scalable RAG/vector architecture. Largest user base on AWS Bedrock at the time.
AI recruiting tool covered by Army Times. Shipped MVP in 12 weeks via Scrum/Agile with a 20-person team.
Group PM for a portfolio of 5+ AI automation products targeting talent and HR workflows — defined cross-product roadmap, prioritized investment, sequenced delivery against value cases.
Owned the Deloitte–Anthropic relationship. Trained 1,000+ practitioners on Claude. Led peer-reviewed LLM evaluation research with UVA — directly applicable to GenAI evaluation strategy in regulated R&D.
Built PwC's AI Factory operating model from scratch — 45+ practitioners across 4 workstreams, 15+ production AI assets in 6 months. Authored the Agentic AI Delivery Standards adopted across a 200+ practitioner practice.
End-to-end agentic automation for enterprise customer operations: contact-driver analytics identifies high-volume drivers → knowledge-article generation closes coverage gaps → tool-use creation equips AI agents to act on workflows → deployment across voice, chat, SMS, and email. Built to absorb 70–95% of repetitive service work.
Practitioner guide on LLM evaluation, optimization, and enterprise-scale deployment. Top 5 in "Machine Theory" within the month of publication.
Read on AmazonQuantum computing for ages 6–8. A whimsical dive through the eyes of a curious cat named Whiskers, making complex concepts accessible to all ages.
Read on AmazonAI literacy for kids. Follow Ella on an exciting journey into the digital world — the message is to go outside and touch grass.
Read on AmazonHow AI agents and multiagent systems automate workflows, improve productivity, and enable self-learning. Authored for Deloitte.
Read ArticleArmy Times coverage of Recruit360 — the AI recruiting tool I led the team that built, shipping in 12 weeks and saving 1M labor hours/year.
Read ArticleCoverage of the AI-driven HR tool I led the team that built, serving 22,000 users across J&J Consumer Health (now Kenvue).
Read ArticleAI-themed metal music because why not. Experience AI-generated brutality and sonic chaos.
ListenI run @PromptHubOfficial — exploring Generative AI, vibe coding, LLM evals, and other tech curiosities.
View All Videos on YouTubeSupervised and funded by Deloitte. Led with Anthropic & UVA. Benchmarking evaluation methodologies for large language models in production — directly applicable to GenAI evaluation strategy in regulated R&D.
344,705 tweets analyzed across pre-COVID, pre-vaccine, and post-vaccine periods. Sentiment, topic mining, and voice analytics reveal inverse correlations between COVID-19 cases and positive Airbnb sentiment. International Journal of Web Based Communities, Vol. 21, 2025.
Workforce exit patterns and survival strategies in the auto repossession industry. Published via SAGE Business Cases.
Case study arguing for a coordinated global vaccination response modeled on the post-WWII Marshall Plan. Published via SAGE Business Cases.
Follow-up case examining the logistical and humanitarian challenges of implementing a Marshall Plan-style vaccination rollout across Africa's least developed nations. SAGE Business Cases, 2023.
Personal Profile Generator & Recommendation Engine — AI system for personalized user profiles and intelligent product recommendations using machine learning. Invented and filed during tenure at Johnson & Johnson Consumer Health (later Kenvue). U.S. Provisional App. No. 63/539,865.
It might become valuable one day. No guarantees.
10 guesses. Win and something good happens. 👀