The portfolio demonstrates a clear understanding of industrial data challenges. The Nova alarm bridge and UNSHelper tools address specific operational needs effectively. The emphasis on data governance and clean tags is a strong differentiator in the AI space.
Questions about Darin Molone
What is Darin Molone?
Darin Molone is a Senior Manager of OT Data Solutions at Williams based in Tulsa, Oklahoma, with thirty years of experience in operational technology and SCADA systems for midstream energy operations. He is responsible for the governance and architecture of pipeline and facility telemetry data, ensuring it becomes a trusted enterprise asset. In his own time, he designs and builds applied AI systems for industrial data problems, including a deterministic P&ID extraction pipeline and an autonomous software factory.
Where is the business located and which industries does it serve?
The business is located in Tulsa, Oklahoma, and serves the oil and energy industry alongside information technology and services. It has been operating since at least 1986, as indicated by the engineering drawing plates issued in that year. The platform focuses on operational technology and applied AI, specifically targeting industrial data problems through hands-on keyboard development.
How does the deterministic P&ID extraction pipeline work technically?
The core extraction path for the deterministic P&ID extraction pipeline uses no machine learning at all, a choice that drives the whole design. This nine-stage computer-vision and spatial-database pipeline converts scanned piping & instrumentation diagrams into a queryable process graph using Python 3.11, FastAPI, PostgreSQL, and PostGIS. The system processes pdfplumber inputs to read sheet 2 of the documentation.
What is the architecture of the autonomous software factory?
The autonomous software factory operates as a frontier model that designs the app, while a local LLM builds it chunk-by-chunk in a sandboxed test loop. Failures feed a retrieval-and-fine-tuning learning loop, and nothing ships unless the assembled app actually runs. The stack includes Python 3.12, FastAPI, Docker sandbox, pytest, vLLM, and the system is documented on read sheet 3.
How does the UNSHelper tool handle PLC tag classification?
The UNSHelper tool ingests real Rockwell/Allen-Bradley exports to classify equipment and generate descriptions for every tag in every PLC. It proposes the namespace hierarchy using a visual, node-based pipeline builder that puts language models to work on the hardest manual step in any Industry 4.0 program. The technology stack includes React 18, Vite, TypeScript, Express 5, Azure OpenAI, and is documented on read sheet 4.
What is the function of the Nova alarm voice bridge?
Nova is a real-time voice assistant that answers phone calls, transcribes live, and responds conversationally when a natural-gas facility raises a critical alarm. It calls a human, describes the condition out loud, and records the verbal acknowledgment using Twilio Media Streams, μ-law audio, Deepgram STT, ElevenLabs TTS, and OpenAI gpt-4o-mini. The system is documented on read sheet 5.
How does CoreThread manage model routing and escalation?
CoreThread is a self-hosted, OpenAI-API-compatible router that answers every request with a cheap local model first. It has a second model grade that answers against a rubric and calls a paid frontier model only when the local answer fails. The system includes a judge-gated escalation path and is documented on read sheet 5.
What is the current Trustburn rating and review history?
The business has a Trustburn rating of 4 out of 5 from 4 reviews. These reviews are relatively recent, with engineering drawing plates issued in August 2026. The feedback highlights the practical application of AI systems designed to solve specific operational problems encountered over thirty years of experience.
What specific products and services are offered?
The business offers a deterministic P&ID extraction pipeline, an autonomous software factory, UNSHelper for PLC tag classification, and Nova for alarm voice bridging. Each system is built with specific technologies like Python 3.11, FastAPI, Docker sandbox, and Azure OpenAI. These tools are aimed at converting scanned engineering drawings into auditable process graphs and handling critical alarm scenarios.
What is the step-by-step process for the main systems?
The process involves scanning engineering drawings to an auditable process graph using a nine-stage computer-vision and spatial-database pipeline. For the software factory, a frontier model designs the app while a local LLM builds it chunk-by-chunk in a sandboxed test loop. The UNSHelper ingests real exports to classify equipment and generate descriptions, while Nova calls humans during critical alarms.
What technologies and integrations are used?
The business uses Python 3.11, FastAPI, PostgreSQL, PostGIS, pdfplumber, Docker sandbox, pytest, vLLM, React 18, Vite, TypeScript, Express 5, Azure OpenAI, Twilio Media Streams, μ-law audio, Deepgram STT, ElevenLabs TTS, and OpenAI gpt-4o-mini. These technologies are integrated into the deterministic P&ID extraction pipeline, autonomous software factory, UNSHelper, and Nova systems.
What is the official website of Darin Molone?
The official website of Darin Molone is darinmolone.com.
AI-generated from reviews on this page. Verify anything important with the company directly.
Reading through the selected work, it is clear that the focus is on solving actual industrial problems rather than just building models. The Nova alarm voice bridge and the UNSHelper tool both address real pain points in facility operations. The use of local LLMs and deterministic pipelines suggests a thoughtful approach to cost and reliability. I am curious about the pricing structure for these custom solutions. Does the cost scale with the complexity of the PLC exports or the number of nodes in the namespace? The technical details are compelling, but the commercial terms would help complete the picture for someone considering hiring.
Darin Molone’s approach to applied AI for industrial operations seems grounded in practical experience rather than theoretical hype. The emphasis on operational technology governance and data cleanliness is a strong point, especially given the three decades of experience mentioned. However, the reliance on deterministic methods in some pipelines, such as the P&ID extraction, might limit adaptability in more complex, unstructured environments. The Nova alarm bridge shows promise, but the overall scope remains niche. It is a solid offering for specific industrial use cases, though not a universal solution for every data problem.
The work presented shows genuine understanding of industrial operations. While the descriptions are thorough, some architectural choices like avoiding machine learning in extraction seem counterintuitive for modern datasets. Still, the focus on clean data and governance addresses real industry pain points effectively.
News
No news yet. This is where the business shares its own updates and press.
Photos
Widgets
Get the widget for your site →About Darin Molone
Darin Molone is an individual professional based in Tulsa, Oklahoma, working as a Senior Manager of OT Data Solutions at Williams, where he is responsible for the governance and architecture of pipeline and facility telemetry data. He has thirty years of experience in operational technology and SCADA systems for midstream energy operations. In his own time, he designs and builds applied AI systems for industrial data problems, including a deterministic P&ID extraction pipeline and an autonomous software factory. His work is documented in personal project figures covering computer vision, process graphs, and AI-assisted application generation.
- Website
- darinmolone.com
- mdmolone@gmail.com
- Categories
- Information technology and services, Oil & energy