The frustration that started it all

I'm a developer and team lead. I've spent years shipping features, putting out fires, mentoring engineers, running demos, navigating reorgs, and learning hard lessons in production. And yet, every single time I sat down to update my resume, I hit the same wall.

A blank page. A blinking cursor. And the vague, sinking realization that I had forgotten most of what I'd done.

Not the big stuff — I could usually remember the headline projects. But the details? The ones that actually made me good at my job? Gone. The time I defused a stakeholder conflict by reframing a deadline. The week I reverse-engineered a legacy system nobody else wanted to touch and saved the team two sprints. The junior dev I mentored through their first production incident. These moments — the ones that show judgment, resilience, and growth — are exactly what interviewers want to hear about, and exactly what vanishes first from memory.

Updating a resume shouldn't feel like an archaeological dig through your own career.

The real problem isn't writing — it's remembering

I started thinking about this differently. The resume itself isn't the hard part. The formatting, the bullet points, the action verbs — that's mechanical work. The genuinely hard part is recall. It's sitting there trying to reconstruct months or years of professional experience from scattered memories, old Slack threads, and the occasional calendar invite.

We live our careers in real time, but we document them in hindsight — if we document them at all. By the time you need to tell your career story, the story has faded.

What if you didn't have to remember?

What if a graph could remember for you?

The core idea was simple: what if every meaningful thing you did at work — every project shipped, every skill demonstrated, every lesson learned, every collaboration that mattered — was captured as it happened, stored in a structured graph, and connected to everything else you've ever done?

Not a wall of text. A graph. Nodes and edges. Your career as an interconnected knowledge base where "that time I led the API migration" connects to "stakeholder management" connects to "technical leadership" connects to three specific bullet points that prove it.

With a graph like that, updating a resume stops being a memory exercise. You'd have a living career memory that you could query. Applying for a role that values cross-functional leadership? The graph already knows which experiences demonstrate it, when they happened, and what the outcomes were. You'd never start from scratch again.

The idea lived in my head for a while. I knew it could work. I just needed the right push to actually build it.

The push: a hackathon with perfect timing

Then Google announced the Gemini Live Agent Challenge on Devpost.

The timing was almost absurd. Here was a hackathon specifically focused on building live, voice-powered AI agents using the Gemini Live API — and I had this half-formed idea about a career companion that interviews you about your week. The two concepts snapped together immediately.

What if the way you capture career memories isn't typing them into a form? What if it's just... talking? Having a real-time voice conversation with an AI coach who asks you what you shipped this week, what challenged you, what you learned. A weekly check-in that feels like catching up with a mentor, not filling out a performance review.

This is where Gemini Live changed everything for me. I'd experimented with text-based chatbots for career coaching before, and they always felt transactional — like filling out a form with extra steps. But a voice agent that can listen, interrupt, ask follow-ups, and handle the messy way humans actually talk? That's a fundamentally different experience. You can hear the difference in how people respond. They open up. They ramble. They surprise themselves with what they remember. Voice drops the filter that typing imposes.

The hackathon gave me the impulse I needed. Not just to build the graph — but to build the interface that makes the graph worth having.

Building it: one developer, a lot of AI

I want to be transparent about something: I built My Job Odyssey as a solo developer, and I leaned heavily on AI tools to do it.

Claude Code for architecture decisions, code generation, debugging, and refactoring. Gemini for the voice pipeline and the AI coaching layer. The entire stack — Firebase Cloud Functions, Cloud Firestore, React frontend, real-time PCM audio streaming via AudioWorklet, career memory extraction — was built with an AI pair programmer at my side.

This isn't a confession. It's the point. The same technology that powers the career coach inside My Job Odyssey is the technology that helped me build it. A solo developer, working nights and weekends, shipping a full-stack voice AI platform on Google Cloud in weeks instead of months. That's what these tools make possible now.

The Google Cloud ecosystem made the architecture surprisingly clean. Firebase handles auth, hosting, and serverless functions. Firestore stores the career graph. The Gemini Live API streams bidirectional audio in real time. Every piece of the stack is a Google service, and they fit together without much glue code. When your auth system, your database, your hosting, and your AI model all live in the same ecosystem, the integration tax drops to almost zero.

What emerged

What started as "I wish I could remember my career" became something bigger than a resume tool:

Weekly voice check-ins where an AI named Alex interviews you about your week — what you shipped, what blocked you, what you're proud of. Alex doesn't just listen — it interrupts with follow-up questions, references your previous sessions, and notices patterns across weeks. It's the career conversation most people never have, powered by Gemini Live's real-time audio and barge-in support.

A living career memory graph that grows with every session. Skills, projects, roles, achievements — all connected, all searchable, all structured. After each voice session, a Cloud Function extracts structured STAR-format career events from the transcript and writes them back to Firestore. Not a document you have to maintain. A graph that maintains itself.

A voice resume editor where you talk to Alex while watching your resume update in real time. This was the feature that made people's eyes go wide. Alex uses Gemini function calling — seven declared tools — to add certifications, rewrite bullet points, update your summary, all while maintaining a natural conversation. Green highlights for additions, yellow for rewrites, amber for flags.

A resume tuner that takes a job posting and tailors your resume using your career graph. It knows what you've done. It scores how well each experience maps to the job's requirements. It rewrites bullets to maximize relevance without losing your voice.

Interview practice scoped to the gap between your profile and the role you're applying for, so you're never blindsided by "tell me about a time when..." again.

Why it's called My Job Odyssey

Because a career isn't a ladder. It's a journey — full of detours, discoveries, setbacks, and surprises. An odyssey. And like any odyssey, the value isn't just in the destination. It's in what you learn along the way.

The tragedy is that most people forget their own odyssey as they're living it. My Job Odyssey exists to make sure you don't.

---

My Job Odyssey is live at myjobodyssey.com. It was born from frustration, shaped by a hackathon, and built entirely on Google Cloud with the Gemini Live API at its core.