Who I Am
Former music educator turned developer based in the DFW area. After realizing the classroom wasn't quite the right stage, I transitioned into tech—starting in support, moving through project management, and eventually finding my home in web development. Over the past 10 years, I've built a career around creating intuitive, high-performing digital experiences, collaborating with cross-functional teams, and mentoring developers along the way. I'm passionate about the intersection of design and engineering—the part where creative vision meets clean, scalable code. Let's build something great together.
Who I've worked with
Over the past decade, I’ve built and led frontend and fullstack solutions for a wide range of organizations—from major corporations like Capital One, Southwest Airlines, and AT&T, to agile teams at Amdocs Studios and fintech startups like LeafHouse Financial.
What I Do
My work has spanned designing modern UI systems, developing scalable component libraries, optimizing legacy codebases, and collaborating across disciplines to ship responsive, user-focused applications. Whether launching internal platforms for auto loans or revamping customer-facing experiences, I’ve consistently delivered clean, performant code that bridges the gap between design and development.
Projects
Flux Barge-In
A full-duplex voice agent built on Deepgram Flux STT + TTS
Most voice-agent demos stop at “text-to-speech works.” This one is built around the part Deepgram Flux is actually designed to solve: what happens to the conversation when a user talks over the agent mid-sentence. Interrupting the agent doesn't just stop the audio — it reconciles the transcript so the model only ever remembers the words the user actually heard, not the sentence it was cut off from saying.
Barge-in with a real playback clock
Flux TTS streams faster than real time, so interrupting is timed against rendered audio frames, not the send clock — otherwise history records words the user never heard.
Live voice & turn-taking control
36 Flux TTS voices, switched live mid-conversation, plus end-of-turn confidence and silence timeout tuned in real time via streaming Configure messages.
Voice interview practice mode
Asks real questions out loud from a rubric-graded question bank and follows up on what your answer actually missed, entirely hands-free.
Resilient by design
Session resumption survives a dropped socket, and the whole pipeline degrades gracefully to a scripted responder with no LLM key at all.