InnoSpark'25 — Hosting the Gemini AI Workshop
Facilitated a hands-on Gemini API session at GDG OC GU's InnoSpark event — exploring prompt engineering, multimodal inputs, and live coding with Google's latest models.
The Event
InnoSpark'25 was GDG On Campus Galgotias University's flagship AI exploration day — a morning of hands-on sessions, live demos, and technical deep-dives for students curious about where AI is headed.
I co-facilitated the Gemini AI Workshop, which ran for 90 minutes and walked through Google's Gemini API from the ground up.
What We Covered
Getting Started — Setting up API keys, installing the SDK, and making the first call. We made sure everyone had working code before moving to the next section.
Prompt Engineering — The difference between a prompt that works and one that doesn't. We explored zero-shot vs. few-shot, chain-of-thought prompting, and system instructions.
Multimodal Inputs — This is where things got interesting. Gemini's ability to accept image + text together opened up use cases participants hadn't considered — captioning, visual Q&A, document parsing.
Live Coding — Built a simple document summarizer end-to-end. Everyone built their own version.
What I Took Away
Running a live coding workshop forces clarity. You can't hand-wave through the parts you're shaky on — you either know it or the terminal tells you otherwise. It pushed me to understand the Gemini SDK more deeply than reading docs alone ever would.
The questions from participants were some of the sharpest parts. One student asked about rate limits in production. Another asked how Gemini handles ambiguous images. Good sessions make you smarter about the gaps in your own knowledge.