This page is the interactive demo: a small language model (Qwen 2.5, 0.5B) runs entirely in your browser and answers questions about my work, and recruiters can run a fit check against a job description. Nothing you type leaves this tab.
I'm Ram Yadlapalli. I lead a 30 person product and engineering organization in Dallas, and I still write code. In 2026 I built an MCP server for customer reporting, a shipment tracking product, and an appointment scheduling agent. Ask about any of it below.
This page runs an open source language model, Qwen 2.5 (0.5B, Apache 2.0), entirely in your browser with WebLLM. Answers start from the page text while the model loads in the background (about 300 MB, once). When it is ready, the model writes the answers itself, using only the profile on this page. There is no server and no API key, and your questions do not leave this tab.
Hiring for a role? Paste the job description. The model compares it with my profile on your machine and writes a fit brief: where the profile matches, where it does not, and an opening line.
Built in October 2026. Customers ask their operational questions through an AI assistant instead of waiting for custom reports. They use it for day to day reporting.
Built in July 2026, in pilot with two customers. An ML model recommends the best appointment for each load. Agents book it by voice on phone calls, by text and email, and through a Claude browser use agent on consignee portals that have no API.
Built in April 2026. Carriers and their customers track a shipment from pickup to delivery, with ETAs, live driver location, and PO and BOL lookups.
How We Built an MCP Server on Our Existing APIs
A production MCP server built on the reporting APIs, identity provider, and Kubernetes cluster we already had. OAuth, multi-tenancy, tool design, and performance.
The platform is a First Mile and Last Mile Optimization and execution product for LTL carriers: a React and TypeScript planner and dispatcher web app, a React Native driver app, Node.js and .NET backends, Azure SQL, and Kubernetes.
I led its 2018 rebuild from an early planning tool into a cloud native platform. It is in production at 11 LTL carriers in the US and Canada, supporting nearly 20,000 drivers and about 1,000 planners and dispatchers, at roughly a million transactions per hour. I am the second highest contributor in its 11 year history, with about 4,220 commits.
Releases are monthly.
VP of Product Development and Engineering in Dallas. I lead a 30 person organization through engineering managers, and I own engineering, the product roadmap, and the business unit P&L.
The product work starts with customers. During the build and rollout years I spent about half my time onsite at carrier terminals and riding with drivers. I sit with planners, dispatchers, drivers, and dock teams, listen to the problem in their words, and turn it into software. I stay in the architecture and the code.
Earlier at MAQ Software I built applications for Microsoft, Starbucks, and Amazon, and led a team of 10 from 2015.
Ram Yadlapalli · Dallas, Texas · 2026
One HTML file, no tracking. The model on this page runs entirely in your browser.