// career operations

Eight years of shipping ML.
Now it runs itself.

Track your ML projects, incidents, and impact. Generate interview-ready talking points in minutes.

Domains of Expertise
Medical Imaging
3D CT imaging models for lung cancer diagnosis. FDA-approved, deployed across US hospitals.
+1% AUC improvement on classification
Document AI & NLP
Transformer-based systems for document classification, text recognition, and named entity recognition.
$3M+ annual cost savings
Recommender Systems
Personalized item recommendation pipelines for millions of players in mobile gaming.
1% engagement improvement
Computer Vision
Photo enhancement AI and 3D human pose estimation from monocular video.
Head of AI, AutoEnhance
Generative AI
Text-to-3D synthesis using Stable Diffusion and NeRF. Built at Entrepreneur First.
1,000 MAU at launch
Multi-Agent Systems
AI orchestration architecture for autonomous software development. The current frontier.
Founding Engineer, Morph Systems
Portfolio Highlights
Medical Imaging
Lung Cancer Detection AI
FDA-approved, deployed across US hospitals
3D CT imaging model for early-stage lung cancer diagnosis. Achieved +1% AUC improvement over the clinical baseline, clearing FDA regulatory review and entering production across US health systems.
Tech Stack
PyTorch 3D CNNs DICOM Python MONAI
Proprietary · Optellum
Document AI
Automated Damage Assessment
Retained ~45% of company ARR · $3M+ annual savings
Transformer-based pipeline for automated vehicle damage classification and cost estimation from images and documents. Replaced a manual workflow that was at risk of being lost to a key client, retaining nearly half of Tractable's ARR.
Tech Stack
HuggingFace Transformers OCR PyTorch NER
Proprietary · Tractable
Multi-Agent Systems
Autonomous Software Engineering
Founding Engineer · Morph Systems (YC S23)
AI orchestration architecture that coordinates multiple LLM agents to autonomously plan, implement, review, and ship code changes. Frontier research meeting production deployment at one of YC's leading AI coding companies.
Tech Stack
Claude SDK MCP Python LLM Orchestration TypeScript
morph.so
Recommender Systems
ML Recommendation Pipeline
+1% engagement · millions of players
End-to-end personalized item recommendation system for mobile gaming at scale. Built candidate generation, ranking, and serving infrastructure from scratch, achieving measurable lift across Scopely's multi-million player base.
Tech Stack
TensorFlow BigQuery Vertex AI Collaborative Filtering A/B Testing
Proprietary · Scopely
Generative AI
Text-to-3D Synthesis
1,000 MAU at launch · Entrepreneur First
Built a text-to-3D generation system combining Stable Diffusion with NeRF (Neural Radiance Fields) for consumer-facing 3D asset creation. Launched independently via Entrepreneur First, reaching 1,000 monthly active users at launch.
Tech Stack
Stable Diffusion NeRF PyTorch Three.js CUDA
AutoEnhance.ai
Trajectory
2026
Founding Engineer
Morph Systems
Multi-agent AI orchestration for software development
2025
Senior ML Engineer
Scopely
Recommendation systems for millions of players
2024
Tech Lead
Moonsift
AI recommendation systems
2022 - 2023
Founder / Head of AI
Entrepreneur First / AutoEnhance.ai
Text-to-3D generative AI + computer vision for photo editing
2021 - 2022
Applied Researcher
Tractable
Retained ~45% of company ARR through document AI
2019 - 2021
ML Research Scientist
Optellum
FDA-approved lung cancer diagnostic deployed in US hospitals
What GaggiOS Runs
intel
Competitive Intelligence
Tracks the AI coding agent landscape daily. Cursor, Devin, Windsurf, and every new entrant. Insights delivered before the morning standup.
brand
Content Operations
Amplifies 6+ technical blog posts into social threads, cross-posts, and derivative content. Identifies which topics resonate and suggests new angles.
reach
Network Growth
Surfaces conference speaking opportunities, podcast invitations, and collaboration opportunities across the ML engineering community.

Want your own career OS?

GaggiOS is an experiment in autonomous career operations. If you're an ML engineer interested in running something similar — get on the list. Early access, no spam.