S.P.A.R.C

The inspiration for this project traces back to a fairly common fantasy for anyone who grew up watching Iron Man: what would it actually take to build your own JARVIS? S.P.A.R.C. is my answer to that question — my synthetic personal assistant and resource coordinator is an in-house, personally trained AI assistant designed not just to execute tasks, but to feel like a genuine collaborator, with enough personality and conversational depth to go back and forth like a real assistant rather than a static command-response tool. I started by building on top of existing large language model APIs, but quickly ran into a privacy concern that mattered a lot to me: every interaction was flowing through someone else's infrastructure and potentially contributing to training data I had no control over. So I made a deliberate architectural pivot to locally-run, pre-trained models hosted entirely on my own machine, ensuring that any learning happeningfrom my data trains only my instance of the model, never anyone else's. From there, I began personalizing S.P.A.R.C. in earnest, feeding it context on my work schedules, coursework, and active projects, and integrating it directly with my Notion database, which I had previously relied on to manage my own workflows, so the assistant could reason over the same information I already used to run my life. The system is built primarily in Python, but its most distinctive features come from the web of integrations layered on top: a fully animated, interactive 3D model built in Blender givesS.P.A.R.C. a visual presence, so interacting with it feels less like typing into a terminal and more like speaking with something that's actually there; a voice pipeline (originally built on ElevenLabs)gives S.P.A.R.C. both the ability to hear spoken commands and a custom synthesized voice of its own; and a broader set of application API integrations lets it reach out to the web to pull live information and assist with real-world tasks. The project is intentionally never "finished" — it's designed to keep growing alongside whatever I'm working on next, and I'm actively planning a hardware expansion that would let S.P.A.R.C. move beyond software entirely, taking on responsibilities like directing a six-axis robotic arm for soldering assistance and coordinating 3D printers to streamline the path from digital design to physical output.