Docs for LLMs
A simple curated list of documents that can be used by LLMs to improve quality on various coding tasks.
Building ML systems and tools that actually ship
I am an AI systems engineer with 7+ years of experience shipping OCR, computer vision, NLP, and LLM-powered products from prototype to production.
These days I work on human-plus-agent delivery workflows, production LLM systems, and the engineering discipline required to scale AI-assisted product development beyond toy demos.
This site is where I document technical experiments, field notes, and hard-earned lessons from building AI products across research, mobile automation, and cloud delivery.
Projects
TranslateX
An iOS-first, minimal UI/UX app that helps interpreters translate between English and Japanese in real time, keeping the interface out of the way so the conversation can stay in flow.
Focus Areas
A quick way into the strongest topics on the site.
LLM Systems
Essays and experiments on long context, token compression, agent workflows, and production LLM usage.
OCR + Computer Vision
Work on mobile UI understanding, OCR benchmarks, and ML features that shipped into testing products.
Product Delivery
Notes from building with AI, choosing stacks, and keeping quality high while shipping faster.
Writings
- Gaining 4 Years of Software Engineering Experience in 4 Months New
- From Vibe Coding to Compound Engineering
- Lessons Learned: Vibe Coding an iOS App from Scratch
- MLUI Mobile: Autify OCR vs. Google OCR
- Token Compression: Reducing Attention Waste?
- Long Pythia
- Machine Learning Features in Autify for Mobile
- Solving Automated App Navigation: A Use-case
- Applying Modern Deep Learning in Autify
- Getting the Most Out of Pre-trained Models
- Recent Advancements in AI
- This Icon Does Not Exist — GAN for Icon Generation
- Deploy ML on Cloud Run
- Cloud Run — Future Tech
- Deep Learning in Cloud