Portfolio research programme / 2024–2026
Senior Program Manager · Azure Commerce · Agentic AI
I connect business, engineering, live-site operations, and sales engineering to improve Azure Commerce. My work spans service architecture, agentic AI, telemetry, capacity, billing, and global delivery.
Swipe horizontally to inspect the architecture
Career as a Transformer: customer, billing, subscription, capacity, telemetry, and risk signals become operating patterns; Azure Commerce architecture and agentic learning loops turn that context into a unified, resilient transactional experience.
The published dossiers show challenge, approach, an illustrative model, and contribution. Reserved entries keep the portfolio structure ready without overstating unfinished work.
GitHub Trending, Hugging Face's trending models, and Hugging Face Daily Papers are discovery signals, not quality scores. Items enter this digest when they combine credible provenance, reusable ideas, active attention, and potential relevance across multiple sectors. Business news is a separate, manually-refreshed signal built from a live, cited web search rather than a free public API — see the note below the tabs for why.
Updated daily
GitHub reports 1,542 stars today. That momentum makes this repository worth tracking for reusable technical patterns and signs of durable adoption.
8,444 total stars · Reviewed 23 Sept 2026CLI tool for configuring and monitoring Claude Code
A framework for building agentic apps
An agentic skills framework & software development methodology that works.
AI security radar
Scans agent skills for prompt injection, malicious behavior, data exfiltration, and supply-chain risk before installation.
A CLI and TypeScript SDK for finding, validating, and fixing vulnerabilities in application code.
An enterprise platform for running and managing MCP servers, addressing isolation and control around agent-accessible tools.
Open-source evaluation and testing for LLM applications and agents, including red-team and trustworthiness workflows.
Primary-source watch
New tools and guidance for securing AI agents and DevSecOps environments through a Zero Trust strategy.
Read the security brief ↗ Microsoft / 27 Jul 2026Enhancing AI security through global AI red teamingThe External Red Team Alliance brings researchers together to advance practical AI safety and security testing.
Read the announcement ↗ OWASP / 17 Mar 2026GenAI Security Project expands its frameworksOWASP reports new open resources and broader industry adoption for generative and agentic AI security.
Read the announcement ↗ MITRE / Living standardATLAS tracks the evolving AI threat landscapeA real-world knowledge base spanning predictive, generative, agentic, and enterprise AI systems.
16 tactics · 178 techniques · 37 mitigations ↗Video world models enable interactive exploration of dynamic environments, yet struggle to respect prior observations over long horizons and across viewpoints. We present WorldCrafter, a video world model that learns a camera-queryable implicit 3D-aware memory…
Modern video games provide a measurable testbed for AI models, combining abilities of visual understanding, instruction decomposition, goal planning, and precise action control over multiple temporal horizons. Existing datasets and benchmarks, however, either…
A multi-agent framework using large language models for stock trading simulates real-world trading firms, improving performance metrics like cumulative returns and Sharpe ratio.
SmolDocling is a compact vision-language model that performs end-to-end document conversion with robust performance across various document types using 256M parameters and a new markup format.
Apodex 1.1 improves sustained, verifiable progress on complex real-world tasks by scaling executable environments and training agents to coordinate long-horizon work with state maintenance and recovery.
YuE, a family of open foundation models based on LLaMA2, can generate long-form music with aligned lyrics, coherent structure, and appropriate accompaniment using innovative techniques in next-token prediction, conditioning, and pre-training.
As coding agents move from supervised code completion to unattended, around-the-clock exploration, their work expands from isolated predictions into long trajectories of reasoning, tool use, and feedback. Token efficiency therefore becomes important for…
OpenDevin is a platform for developing AI agents that interact with the world by writing code, using command lines, and browsing the web, with support for multiple agents and evaluation benchmarks.
An LLM agent's capability is largely magnified by its harness, namely the prompts, control flow, tooling, memory, and context management surrounding the frozen backbone model. Recent methods increasingly automate this process by iteratively proposing and…
FreeToken is an edge-native Mixture-of-Experts serving system that dynamically maps computation and model state onto heterogeneous local hardware to run large open-weight models on personal machines.
convaiinnovations · Text Classification
A text classification model built with transformers, published by convaiinnovations on Hugging Face.
Hugging Face reports 2,906 likes and 0 downloads. That combination of interest and adoption makes this release worth tracking for capability shifts and new default choices.
0 downloads · Reviewed 23 Sept 2026A text to image model built with diffusers, published by Qwen.
A text generation model built with llama.cpp, published by prism-ml.
A text generation model built with transformers, published by XingChen-AGI.
News radar
The AI security startup raised a $100 million round as enterprises push to harden their AI deployments.
AIR raised $50 million to help companies vet the skills and add-ons used by AI agents.
OpenAI unveiled an initiative aimed at helping utilities defend against AI-enabled cyberattacks.
A new House bill would establish security standards for companies deploying AI agents after the Hugging Face hack.
Also worth tracking: The business news tab above is refreshed manually, not on the same daily schedule as the signals above — it calls a paid Foundry web-search tool, so it runs by hand every so often rather than automatically. For continuous, comprehensive coverage of AI investment and global datacenter buildout, here are two reputable, actively maintained trackers instead: Crunchbase News: AI ↗ for funding, and Data Center Dynamics ↗ for global buildout.
A career connecting platform architecture with the financial, operational, and human systems required to run cloud services globally.
Sep 2021–Present
Blend business, engineering, live-site, and sales-engineering perspectives to improve and unify transactional experiences across Azure Commerce.
Lead commerce service architecture reviews, AI feature engineering, and continuous improvement through agentic harness and loop engineering.
Dec 2017–Sep 2021
Helped enterprises convert complex Azure commerce, billing, identity, and capacity challenges into predictable executive outcomes through rigorous technical and economic analysis.
Built telemetry frameworks for decisive action, improved operational resilience, and launched delivery centers in Costa Rica and Colorado Springs supporting more than 1,000 engineers.
Nov 2009–Dec 2017
Resolved complex Azure cloud infrastructure, subscription, and billing issues while translating emerging operational patterns into clear business needs and action plans.
Led small rotational teams and guided projects from incubation through operational maturity.
May 2024–Jan 2026
Shiley-Marcos School of Engineering · University of San DiegoGraduate study reinforcing the technical foundation behind applied AI program design, evaluation, data grounding, and responsible adoption.
The evidence combines graduate study in Applied Artificial Intelligence with a curated selection from 18 public Credly badges.
Microsoft Business Operations
The Microsoft Garage
Pendo
Pendo
Pendo
Microsoft MCAPS Academy
Microsoft technical community
Microsoft
Additional verified work includes Microsoft Global Hackathons from 2020–2025, Product Management Basics, Pendo Super Certified, Kepner Tregoe Practitioner, and Azure Subscription & Billing Support.
Independent AI exploration sits alongside two ways I reset away from a screen: E-skates and snowboarding.
Exploring how persistent AI use may change independent reasoning, creativity, socialization, and decision accountability.
Open research dossier ↗Turning early ideas into testable frameworks, telemetry models, and working prototypes that make abstract questions observable.
Review selected projects ↑Extending practical experience through graduate study, product education, technical communities, and hands-on experimentation.
Review credentials ↑A fast, focused way to get outside, explore, and exchange screen time for movement.
My home mountain is Crystal Mountain in Washington—where winter weekends mean fresh air, elevation, and the next run.
06 / About
This portfolio connects selected work with the operating experience behind it: Azure commerce architecture, customer economics, live-site resilience, telemetry, global delivery, and applied AI.