Metropolis Technologies

The pace of innovation in AI and machine learning isn't just fast; it's exponential. To stay at the forefront, we can't just hire smart people — we have to create a system where the entire team learns together. We have found that a successful learning culture is built on a foundation of consistent, structured knowledge-sharing.

On a daily basis, we use dedicated forums to share new research, open-source projects and industry insights. This isn't just about sharing links; it's about building collective intelligence. Another important practice we embrace is our weekly reading group.

We cover a vast range of topics in a structured theme, from industry best practices and the latest literature in AI research to foundational papers that relate directly to our work and groundbreaking topics like LLMs. The discussion goes beyond a simple review.

We frame everything through the lens of our work, asking, "How does this new technology apply to the challenges we face at Metropolis?" This exercise builds a shared mental model of what's possible and what's next.

A culture of learning isn't a vanity metric; it's a force multiplier for innovation and execution. When the entire team shares a baseline understanding of a technology, you eliminate the psychological barriers and friction that come with adopting it.

For example, after our team collectively studied the latest research on large vision language models, we saw an opportunity to rethink our data annotation pipeline. That shared understanding directly enabled us to bypass weeks/months of traditional exploration and research involving proof-of-concept work.

Instead, we were able to quickly prototype and deploy a custom VLM solution that significantly streamlined our workflow. That wasn't just a technical win; it was a testament to our team's agility, born from that shared knowledge.

"A culture of learning doesn't happen by accident; we have to be intentional. " Learning and educational sharing should be celebrated and expected, not tolerated." " Lead by example.

We must be the biggest learner in the room. Be the first to bring a new paper to the reading group or share an interesting finding. Let the team follow your lead.

" " Align incentives. Learning takes time and energy, and that has to be a recognized investment. Whether it's a small prize for the best presentation, a dedicated learning budget or including it in performance reviews, you have to show that you truly value it.

Without that alignment, it will be the first thing to go when deadlines loom." Ultimately, a learning culture isn't just a benefit; it is the fundamental foundation of a high-performing engineering organization.".

From day one, you are exposed to exciting analytical projects and implementations. I'm part of an environment where personal growth and clarity are as important as challenging the status quo in mobility.

At Metropolis, we're an AI company for the real world. Our AI-powered recognition platform and effortless payment system have transformed one of the most analog industries — parking. We replaced paper tickets and cash with a seamless drive-in, drive-out experience.

Now we're expanding to new real-world interactions like refueling, drive-thrus, retail and stadiums. Our engineering teams focus on building and scaling technology across 4,200-plus sites serving more than 50 million customers. Key priorities include advancing our proprietary vision systems — Orion and BigMac — creating personalized, frictionless payment experiences, ensuring reliability and security at scale and supporting partners across parking lots, airports and cities through robust data integration and analytics.

Recently, we've turned our AI focus inward to tackle a real-world challenge: building Metropolis itself. This led to our newest initiative — context engineering — a service designed to manage AI systems by giving them the right information and tools at the right time.

When we introduced AI tooling our Visit Enablement team took a fresh project off the roadmap to improve our map search capability with clustering capabilities. Instead of many labels on a zoomed out map we would show a cluster with the number of sites.

Our normal estimate for a project of this size with frontend and backend work was two to three weeks. An "All AI" approach using tools like Google Gemini, Github Copilot and Claude Code allowed the team to deliver a working prototype in less than a week.

They were able to quickly iterate on the product requirements, the visual design and the frontend code/logic using MCP tools. The backend team also developed better search capabilities and used a common API to connect the systems.

Without our new AI capabilities, the map clustering project would have faced multiple transitions and delays from team handoffs. A project of this nature would also have required significant time from our busy frontend team. We've taken a holistic approach to AI tooling — starting with a tinkering phase that yielded early gains in code completion and reviews.

Once agentic AI became widely available in May, we went all-in, using systems thinking to automate entire problems instead of isolated tasks. Our AI transformation came in three areas: developer acceleration — faster, higher-quality code, tests and documentation.

Parallel execution — developers can offload security reviews, flaky test fixes and meeting agendas to AI. Process elimination — with MCP services we go from visual design to frontend code and with spec-driven development from idea to implementation, reducing handoffs and delays.

We're now building an "AI Developer" program, managing evolving tools, tracking ROI and investing in context engineering to make AI and humans more efficient. Despite added training, productivity gains are significant, creating a virtuous cycle that aligns with our mission to make the real world.

The engineers that are successful on our team know how to focus on not only the technical implementation but also have the ability to deliver results that match the product vision and business needs. Ultimately the more the individual contributes to team success the more they will achieve individual success.

Building a network of technology-enabled real estate. We use machine vision, IoT, apps, and distributed computing to blur the boundary of the physical and digital world. Every day is a challenge and a ton of fun.