Writing code is what I do. References available upon request. Please reach out through my contact form.

Brief synopses of some of our projects:

AI Content Generation is quickly becoming one of the most common projects we’re seeing requested. In one such system for an API integration for developers, we couple code with support requests to make AI-suggested improvements to the developer documentation. Another simpler example is for creating customized exercise plans based on the user’s specific goals and needs to help them build good habits. We’ve also built AI generated content for children’s educational platforms: one where we generate compelling stories one word at a time in a typing adventure game, and another where we combine text and imagery as part of controlled linear experiences in an ed-tech mobile app.

Custom RAG solutions are the next evolution in AI content generation, combining embeddings from an LLM to find the most semantically-relevant documents to include in the context of a prompt to generate more text. You can think of a RAG as being like search on steroids, where instead of finding the most similar documents based on specific words it finds them based on semantic relationships. These are often used in tools like Onyx (formerly Danswer) to allow users to chat with a huge set of documents, but can be applied to any problem combining AI with large bodies of text.

MCP Servers can be thought of as a general interface to any specialized API and are quickly gaining traction with developers using agentic tools and beyond. The way a human might use a web browser to access a website, an AI agent can use an MCP server to access an API. This is quite open-ended but there are already a number of practical uses being seen like in developer documentation. Since so many developers are using agentic tools in their development process now many larger companies are starting to embrace MCP servers and make their documentation available through them. This lets the AI agents use the documentation more effectively as they use as less and more focused text. MCP servers are even being used to expose access to logs so that agents can retrieve immediate feedback without human assistance.

Saas Products are frequently the next step for companies that have developed an internal custom software that is serving them well. By building a SaaS product they can take their knowledge and process and license access to it or use it as a competitive advantage in their field. We also build SaaS products for people with ideas that want to get them out there.

Support and Maintenance is needed as long as an application is in use. The environment that applications live in is always changing with new threats evolving daily which means systems need to be evolve and stay up to date. We regularly help companies take over legacy projects with attractive support rates and 24/7 SLAs.

Web Portals/Dashboards are often used by companies to establish a single “source of truth” for their data, outside of the walled gardens that desire to contain it. The web portal/dashboard becomes the main interaction point for employees and customers, replacing interactions with vendor-based software. By controlling their data, our clients can negotiate with their vendors from a position of strength.

Data Scraping is a fantastic source of leads for companies in a number of fields, but is also an ever-evolving game of cat and mouse. Especially today with AI bots scraping everything, there are more and more protections being put in place to thwart would-be scrapers. We use proprietary technology evolved from 20 years of scraping data to defeat anti-scraping measures and get you the data that you need. Coupled with a custom portal or integrated into your CRM you can have leads flowing in daily.

Autoscraper is a web-scale scraping system that I built which leverages AI agents to discover sources, evaluate the available data, and write/maintain the code that does the extraction and parsing. For example we are able to simultaneously and effortlessly scrape public data from hundreds of public US court websites. The agents do it all, including updating the scrapers/parsers code automatically to adapt to changes in target websites. The code integrates into a proprietary orchestration service that distributes requests across thousands of servers and stays largely undetected. If you want the data we will help you get it, without regard for terms of service and other soft restrictions.