# Seed Forty Two > Applied AI, Hybrid Infrastructure and Embedded Systems. We build solutions that span AI models, embedded hardware, and infrastructure. From physical prototypes to distributed deployments, we combine these disciplines to meet the performance and reliability the work requires. The links below provide Markdown versions of the public website. They are generated from the same source content as the HTML pages. Each page links to its original website URL. ## Website - [Homepage](https://www.seedfortytwo.com/index.md): Services, case study summaries, company approach, and contact information. ## Case studies - [Training AI models to read documents accurately in seconds on local hardware](https://www.seedfortytwo.com/work/training-ocr/index.md): We trained and optimized AI models for accurate document and product-label recognition in seconds on NVIDIA Jetson AGX Thor. We adapted training and validation methods from a larger mixture-of-experts model to a compact 1B-class model, with runtime tuning for each. The goal was to keep local deployment costs low while retaining control of the adapted models and their serving runtimes. - [Overcoming the challenges of day-to-day export compliance checks](https://www.seedfortytwo.com/work/manufacturer-eccn-data/index.md): We connected product classification, destination controls, and regulatory explanations in one workflow, making routine export research less dependent on specialist knowledge and repeated searches across separate sources. - [Running a neural audio model in real time on low-power, cost-effective embedded hardware](https://www.seedfortytwo.com/work/embedded-neural-audio/index.md): Limited power from connected equipment and the cost of large deployments shaped this embedded noise suppression system. We implemented a neural audio model for speech noise suppression on the selected hardware and retained its behavior in reference comparisons. The integrated audio configuration processed frames within a 16 ms budget. - [Correcting a severe hidden scaling defect in 4-bit quantized AI inference](https://www.seedfortytwo.com/work/inference-correctness/index.md): 4-bit quantization can reduce memory use. We found missing scale factors that severely distorted an expert-network calculation. In a controlled component test, a host-side correction brought the output close to the runtime’s quantized reference. ## Privacy - [Privacy policy](https://www.seedfortytwo.com/privacy/index.md): How Seed Forty Two LLC handles website enquiries, security cookies, and website analytics. ## Complete content - [Full website content](https://www.seedfortytwo.com/llms-full.txt): The homepage, all 4 case studies, and privacy policy in one file.