Training and Fuelling for Female Physiology
Seven places default training advice gets it wrong for active women, and what to do instead.
Mulloy Morro
Founder. Tech Leader. Builder.
15+ years building AI/ML systems from research to global production, spanning consumer tech, healthcare, legal tech, and environmental applications. Former Spotify ML leader. Now founding human-centered AI products.
Writing
What I’m writing
Long-form essays on the mind, longevity, and how to train — plus notes on recommendation systems and building.
The longevity diet, decoded
What we eat tunes the cellular machinery that determines how quickly we age. A practical breakdown of diet patterns, protein quality, methionine load, and evidence-backed plate changes.
The Body Beneath the Feeling
Where do emotions come from, and what lies beneath awareness? A guided tour of the rival answers — ending in a staged debate among five thinkers who would not agree about any of it.
Experience
A journey through AI/ML leadership and product development
Nalya.ai
Founder & AI Architect
There's a quiet gap between two categories of apps. The know-yourself tools — Headspace, Hinge, BetterHelp — help you understand who you are. The activity marketplaces — Playtomic, PlayByPoint, Resy, OpenTable — help you book a court or a table. Neither helps you find your people and actually show up together. A Sunday evening, an empty week ahead, a padel court two taps away and no one to play with — that's the gap. Nalya flips it. By default, you wake up Monday with three padel matches already on your calendar, matched to your level, your schedule, and your vibe. Your only job is to opt out if something comes up, and we find a replacement so the other three still play. When showing up is the easy thing, a rhythm forms. When the rhythm forms, the community grows, the courts fill, and the loneliness number starts moving the other way. The know-yourself layer and the activity layer finally talk to each other.
Lumi AI Agent Two surfaces, one loop: Discover finds your people; Lumi makes the match happen.
Lumi AI Activity Agent · via SMS
Active
Scroll into view to start
Legal-Tech Joint Venture
Technical Co-founder, AI & Systems Lead
Led an AI-enabled remediation initiative applying machine learning, LLM-based retrieval systems, and systems thinking to environmental damage claims related to PFAS contamination in drinking water. Led fundraising efforts securing $4M in funding, grew team from 1 to 15 FTEs, achieving $80M valuation within 6 months.
Spotify
IC → Engineering Manager → Group Lead
Grew from founding engineer to group lead over 4 years, driving Spotify's personalization strategy from early-stage R&D through company-wide adoption. Led the Home Construction organization (25+ engineers) owning recommendation systems that power the homepage where 95% of consumption begins for 600M+ MAU.
Compass Real Estate
Staff Machine Learning Engineer / Founding MLE
Built early ML infrastructure and applied models supporting growth and personalization. Founded AI teams in New York and Seattle, developed ML platform, created new ML product metrics, and increased MRR and CTR of Compass search.
OfferUp
Founding MLE / Head of Machine Learning
Founding Machine Learning Engineer focused on content understanding models spanning language, vision, and user behavior. Applied ML to search ranking, recommendations, user and content moderation, and automating item-posting flows. Helped scale and structure the Data and ML organizations.
Surgical Robotics Company (Confidential) Side Quest
ML Expert Consultant
Provided strategic ML consulting for a surgical robotics company while at OfferUp, advising on machine learning applications in surgical automation, computer vision for robotic guidance systems, and evaluating emerging technologies for real-time surgical assistance.
Porch Group
Senior Machine Learning Engineer
Built ML systems for the leading home services marketplace connecting homeowners with local service professionals—electricians, plumbers, contractors, and more. Developed unsupervised learning models to understand service professional behavior and optimize matching.
UC San Diego
M.S. Intelligent Systems, Robotics & Control
Graduate research in Computer Vision under Prof. Nuno Vasconcelos. Thesis focused on Generative AI, video segmentation, and person-count regression.
UC San Diego
B.S. Engineering Physics & B.A. Visual Arts, Media
Electrical Engineering and Physics interdisciplinary program combined with creative visual arts and media production.
Projects
Building AI systems that deliver measurable impact
FeaturedSpotify's New Home Feed
Led the redesign and modernization of Spotify's main-feed recommendation system serving 800M+ users. Delivered $10M/month increase in user margins, 26% boost in discovery, and $4M/year reduction in operating costs.
FeaturedAgentic-AI for PFAS-Free Drinking Water
Leveraging agentic-AI to accelerate environmental remediation and deliver cleaner drinking water to affected communities. Designed and implemented a HIPAA/HITECH compliant system for secure medical records processing, ensuring proper handling of protected health information (PHI) while analyzing medical documentation related to environmental exposure claims.
FeaturedSquad Recommendations Engine
How it works
FeaturedLumi — AI Activity Agent
Lumi AI Activity Agent · via SMS
Active
Scroll into view to start
FeaturedNalya.ai
A human-centered agentic-AI consumer platform focused on augmenting real-world human connection through privacy-aware, human-in-the-loop ML systems.
Marketplace ML Platform
Built ML infrastructure and recommendation systems at OfferUp, one of the largest mobile marketplaces in the US. Scaled moderation team efficiency through automated bad actor and prohibited item detection.
Pro Service Area Modeling
Unsupervised learning system to model service areas of home service professionals on Porch.com, inferring coverage zones from sparse job location data, CBSAs, and collaborative filtering of colleague behavior.
FeaturedUser Lifetime Value Prediction
Developed and scaled Spotify's north star metric: User-LTV (lifetime value), adopted company-wide to quantify cumulative causal impact and perform financial scenario analysis on long-term retention. Featured at Investor Day 2022 by Head of ML Tony Jebara.
4 Crowd Understanding
Computer vision platform for automated monitoring and surveillance of crowded scenes using dynamic textures.
Healthcare Fraud Detection Platform
One of the first projects in my career: designed and built a production ML platform using Java and Python for healthcare fraud detection that secured a major contract to build the fraud detection component of a national Federal Healthcare Information Exchange.
🚁
Drone Video Analysis
Cluster video segments into story-telling salient clusters with auto-editing using telemetry data.
Beyond Work
Adventures, music, and the things that inspire me
Adventures Photo Album
Snapshots from travels, hikes, and explorations around the world. Life's best moments captured.
View PhotosMy Spotify
Follow along for curated playlists, podcast recommendations, and what I'm currently listening to.
Follow on SpotifyCurrent Favorites
🧠 Deep Focus
Flow State Radio
Music for deep work & concentration
Hand-curated playlists designed for writers, coders, and builders. No vocals, no distractions — just steady energy to help you lock in.
Start Focusing
Let's Connect
Interested in collaborating, discussing AI/ML, or just saying hello? Book a time that works for you.
Whether you're exploring a new venture, need ML expertise, or want to chat about building AI systems that matter — I'd love to hear from you.
Support My Work
If you've found my projects or content helpful, consider buying me a coffee! Your support helps fuel more open-source work and creative experiments.
Buy Me a Coffee