Senior ML Engineer working on document intelligence and engineering analysis. Previously shipped mobile and cloud computer vision at Carb Manager and developed financial-document recognition at CFT.
Role: Senior ML Engineer Portfolio: https://zack-dev-cm.github.io/ Contact: mailto:kaisenaiko@gmail.com
- Computer vision: OCR, detection, segmentation and multimodal retrieval, from model experiments to server and on-device inference. Examples: Dermaself Flutter Skin Analysis App, Multimodal Video Search Platform, Fast OCR ONNX Inference Server.
- Document and agentic AI: Hybrid retrieval, source-linked answers and human review for document and tool-based workflows. Examples: Document AI, Construction Document Intelligence, InQuest.
- Inference and delivery: Profiling, model packaging, evaluation and deployment across cloud and mobile. Examples: Agnitra - ML Profiling & Optimization, CV Repro Lab Skills.
- Document AI: R&D for source-linked specialist review: document extraction, deterministic checks and retrieval over reference material.
- Architectural Drawing and Interior Catalog Matching: Explore drawing-to-catalog matching, dimension checks and a furnished-room fit in an eight-stage 3D guide. Three recorded source sets show the workflow; live generation is currently paused.
- Vehicle Lab: A Reusable Engineering Notebook: Built and released an engineering notebook connecting CAD inspection, digital prototyping, revision decisions and recorded simulation, with a reusable Python CLI and interactive 3D viewer.
- SectionCheck - Image Registration Review: A CPU prototype for inspecting a proposed affine image transform, reviewing polygon annotations and verifying decision-bound exports on synthetic data.
- Dermaself Flutter Skin Analysis App: Skin-analysis computer vision for Dermaself, connecting guided mobile capture with pore and wrinkle segmentation and usable results.
- Agnitra - ML Profiling & Optimization: A published Python SDK and CLI for inspecting model runtime and applying inference optimizations within existing ML workflows.
- Calorio - AI Nutrition Service: A maintained Telegram service that helps people keep a food diary with meal photos, voice messages and text.
- Multimodal Video Search Platform: Video search case study combining keyframes, ASR/OCR, object and face signals, visual embeddings, transcript embeddings, and hybrid retrieval.
- Resume PDF: Current resume.
- Resume HTML: Experience and project summary.
- Project data: Structured public project information.
- Full project text: Project scope, implementation context and proof links.
- Research notes: Applied ML paper reviews.
For each project, use its canonical HTML page for the case study and its public links for product, package or repository evidence. Preserve the stated scope of R&D and prototype work.