As part of the Google Magi initiative, Google was developing next-generation AI experiences powered by Gemini and Google AI Mode, where response quality, safety, and accuracy were critical before rollout. Early efforts surfaced layered challenges: Validating Gemini and AI Mode responses in real-world conditions, ensuring outputs were factually correct, relevant, and policy-compliant, detecting failure patterns like hallucinations and misleading confidence, missing human-in-the-loop, domain-grounded evaluation, and scaling evaluation across multiple domains and languages. Google Magi needed a structured, expert-driven evaluation framework to establish baseline standards, then scale across specialists, ensuring Gemini outputs could be trusted and continuously improved. The engagement became the first Topcoder-led workstream for Google Magi, rolled out in two phases. In Phase 1, Topcoder translated Google's quality guidelines into clear, repeatable evaluation criteria and assessed Gemini and AI Mode responses across correctness, relevance, reasoning, and safety, establishing baseline standards and surfacing early response risks. In Phase 2, the program scaled into expert-driven evaluation, onboarding a curated team of domain specialists across Sports, Finance, and German language to apply structured rubrics and cultural and linguistic accuracy checks. All evaluation outputs were consolidated, quality-checked, and fed directly back into Google's training pipelines, contributing to model tuning, improved grounding, and better domain-specific response behavior. The Google Magi evaluation program delivered foundational, long-term value across AI quality, scalability, and trust. It improved the accuracy, relevance, and reliability of Gemini and AI Mode responses, enabled early identification of hallucinations and domain-specific weaknesses, and strengthened grounding before wider exposure. The engagement established the first Topcoder-led evaluation framework for Google Magi and created a scalable, expert-driven model reusable across domains and languages. This model reduced rollout risk, increased confidence in AI Mode outputs, and laid the foundation for future multi-domain, multilingual AI evaluation programs.

Skills

Project Details

Wipro's AI practice was scaling rapidly, with growing demand for hands-on Gen AI and Agentic AI talent across delivery, innovation, and AI Scale initiatives. They faced difficulty in objectively validating real, hands-on capabilities and an over-reliance on resumes and interviews that didn't demonstrate solution-building skills. Wipro needed to identify industry-ready AI engineers globally across experience bands within compressed timelines. They sought a scalable, unbiased, and merit-based talent identification model to find top talent for potential full-time roles. Wipro partnered with Topcoder to run two parallel, real-world Codathon challenges, focusing on Generative AI and Agentic AI. These challenges were designed for talent identification and technical evaluation, with participation restricted to India, USA, and Canada, and segmented into 2-4 and 5-8 years experience bands. Topcoder managed the full lifecycle, including challenge design, community outreach, independent technical review, and objective scoring. Winning entries were deployed on AWS for validation, and top performers were awarded prize money and shortlisted for Wipro's hiring pipelines. The Codathon successfully identified 7 AI engineers who were shortlisted for next-stage hiring based on real code and deployed solutions. This model enabled merit-based, evidence-driven talent decisions, reduced hiring risk, and accelerated the AI talent pipeline. It also compressed the full hiring lifecycle from months to weeks. Beyond immediate hiring, the program established a repeatable model for future AI, Data, and Cloud-native engineering needs.

Skills

Project Details

A leading medical device company aimed to enhance the usability and adoption of its mobile application used by clinical personnel. The app was designed to log surgical procedure data following operations involving the company’s devices. However, the existing interface was outdated and unintuitive, making it difficult for field users to capture accurate information efficiently. The objective was to redesign the application to deliver a cleaner, more intuitive experience aligned with iOS standards and better suited for fast-paced clinical environments. Topcoder launched a global design challenge to reimagine the mobile experience. Designers were invited to rethink screen flows, navigation, and information architecture, resulting in modern concepts that prioritized clarity and ease of use. The new app design featured a streamlined login screen, a refreshed dashboard for viewing and filtering procedures, a detailed procedure view with editable fields, and an improved form for creating new procedures. The visual design emphasized readability and simplicity, reducing friction during field use and improving task efficiency. The challenge delivered a complete redesign of the mobile application that significantly improved the user experience for clinical professionals. The updated interface made it easier to log and access surgical data, leading to greater adoption and satisfaction among users. Topcoder’s crowdsourcing approach enabled the client to receive high-quality, innovative design concepts within a short timeframe, offering a faster and more diverse alternative to traditional design methods.

Skills

Project Details

NASA needed improved automated crater detection for analyzing planetary imagery. Existing approaches limited how quickly they could advance image processing methods at scale. They also needed to access a wider range of data science and machine learning expertise beyond internal teams. NASA implemented a competitive marathon match to crowdsource crater detection algorithms. A $49,000 prize pool was used to attract data scientists and machine learning experts from around the world. Participants developed image processing solutions tailored to planetary imagery analysis. The initiative delivered new crater detection algorithms for planetary image processing. The $49,000 prize pool helped draw global participation from top practitioners. NASA expanded the diversity of problem-solving approaches and accelerated scientific research through the competition format.

Skills

Project Details

Top 50 executive or entrepreneur highlighted on the #1 podcast covering the intersection of technology and talent.

METI built a database with thousands of business case studies to help Japanese business owners navigate key decisions and obstacles. The resource went virtually unused because many case studies existed as hard-to-search PDFs. Users often gave up in frustration. Inconsistent formatting and varied content management methods added further friction. METI pursued a more consistent, intuitive, modern digital experience. A design-first exploration and rapid prototyping approach was implemented to produce a responsive website interface optimized for older and elderly users. Designers submitted over a dozen unique interface concepts in less than two weeks, and METI iterated through feedback on preferred options. The effort then delivered a REST API and a fully functional prototype, with all design and development completed in three weeks. The database became more user-friendly and modern, turning it into an essential digital asset for Japan’s economy. The approach delivered design and prototyping 3X faster than traditional delivery options. Total cost was reduced by 125% versus traditional delivery options. METI also received 10+ high-quality design concepts to choose from within two weeks and a full prototype in three weeks.

Skills

Project Details

Stanley Black & Decker aimed to transform collaboration and decision-making across its 150+ factories worldwide. The company needed two connected solutions: a hands-free mobile experience for factory technicians and a centralized command center interface for real-time oversight. The mobile tool needed to help technicians troubleshoot machinery using Augmented Reality glasses while connecting with remote experts. The command center needed to provide live visibility into asset health, production flow, and workforce coordination. A series of three Rapid User Experience (RUX) challenges was launched to conceptualize and refine a visual collaboration platform connecting factory technicians with remote machine experts. Designers created two interfaces: a hands-free Augmented Reality experience for technicians and a touchscreen experience for experts using tablets. The Augmented Reality interface enabled access to live machine data, blueprints, and remote guidance through real-time voice and video via voice control. The expert interface supported monitoring sensor data, reviewing asset history, and collaborating through annotations, remote commands, and live video. Stanley Black & Decker received a forward-looking design foundation to support digital transformation in manufacturing operations. The Augmented Reality tool improved technician efficiency and enabled remote collaboration with experts. The command center enhanced visibility and responsiveness at a global scale. The approach accelerated innovation while keeping user needs at the center.

Skills

Project Details

The U.S. Department of Energy needed to accelerate solar energy adoption by addressing soft costs and key adoption barriers. It required a way to move from ideas to testable concepts quickly. It also aimed to attract broad participation and high-quality solutions. The Department of Energy launched the SunShot Catalyst competition series to crowdsource solutions and prototype development. Finalists received $25,000 each to develop prototypes and MVPs. The program also provided access to an online platform to support development and iteration. The effort delivered 17 solar energy business prototypes. A global community of 165+ members from 61 countries contributed to the work. All work was completed in 9 weeks. The program won the ISPIM Grand Prize for innovation management excellence.

Skills

Project Details

Thompson & Litton needed to update its online presence to keep pace with a fast-growing business. The firm wanted a modern website experience without relying on a traditional one-on-one designer relationship. Stakeholders also needed enough options to confidently choose a direction for the redesign. The team implemented a crowdsourced design competition model to source web and UI design concepts from a global community. Decision makers reviewed close to 20 designs, narrowed the field to 3 finalists, and selected a winning concept. The winning design was then developed and deployed in just under a month. The approach delivered 40% savings versus a traditional agency. Thompson & Litton received 31 design submissions and completed deployment in 26 days. The engagement produced 1 unique, final design and a modern web experience positioned to impress prospects and delight clients.

Skills

Project Details

Land O’ Lakes owned and operated dozens of B2C and B2B brands across dairy, animal nutrition, and crop-focused businesses. Maintaining consistently strong digital experiences across many different web properties created complexity for the centralized digital team. Different brands had different needs, budgets, and governance requirements, which made it difficult to deliver a consistent approach for redesigns and launches. Land O’ Lakes implemented a crowdsourced model to give business units easy access to web design and development talent. The team executed multiple brand website redesigns, implementations, and launches using an on-platform delivery approach. Dedicated Copilot support guided and managed the work and helped organize delivery through the platform. The program included challenges such as responsive UI design and prototyping efforts for Purina Mills merchandising. Land O’ Lakes designed, developed, and launched 5+ unique brand websites. The approach delivered 300% faster time-to-market than traditional means available. Land O’ Lakes also reduced cost per website redesign by 70% compared to agency options. With no lead time on new projects, brand leads were able to start digital work quickly while maintaining confidence in quality and budget fit.

Skills

Project Details

Macy’s needed to run a live design thinking exploration during NRF’s annual “Big Show” while e-commerce innovation was a central focus of the event. Leadership wanted to rethink macys.com (https://macys.com) beyond transactional functionality and better express the brand promise of being a fashion authority. The team also needed multiple modern design directions that could showcase multiple modern looks. The effort required rapid execution and fast turnaround. A crowdsourced LUX (Live User Experience) redesign concepts challenge was planned and executed around the NRF event. The work focused on reimagining key consumer experiences on macys.com (https://macys.com), with goals including personalized outfitting and shifting users away from a coupon-driven mindset toward value tied to style and emotionally rewarding experiences. Concepts explored modern technology opportunities such as AR, AI, and shoppable video. A rolling checkpoint was used to provide in-progress feedback, and an AI-based shopping assistant feature selected by NRF attendees was added to the specification. The effort produced 18 unique UX/UI design concepts for a new macys.com. More than 180 individual screens were delivered back to the Macy’s team from the fast design challenge. The community participation included 41 registrants representing 7 countries. The overall timeline moved from initial idea to final results in under three weeks, with the design challenge itself completed in less than two days.

Skills

Project Details

Comcast aimed to refresh the design of its Careers website to deliver a modern, responsive experience for job seekers while simplifying site maintenance. With over 184,000 employees and thousands of job openings, the design needed to appeal to both internal and external candidates. The effort also needed to streamline user navigation and remain consistent with Comcast’s brand identity. A rapid visual design challenge was implemented to transform key job-related pages. Participants redesigned the Landing Page, Search Jobs, Apply Job, Job by Location, and Job by Title. The entries included improvements in layout, animations, typography, and visual hierarchy. Designers submitted responsive mockups for desktop, tablet, and mobile while adhering to corporate brand guidelines and the existing navigation framework. Comcast benefited from a fresh, engaging, and responsive Careers site aligned with brand standards. The redesign enhanced the job application journey for candidates. It also simplified future updates for internal teams.

Skills

Project Details

T-Mobile needed to deliver a high volume of digital work while facing constraints on full-time staffing and access to hard-to-find skills. The organization also had to support a wide range of initiatives across design, development, and QA. Teams sought a way to increase delivery speed without sacrificing quality. T-Mobile established an ongoing partnership to tap an on-demand, global talent pool and run multiple simultaneous work streams. The work spanned design, development, and QA initiatives across 85 projects. Projects included GitLens enhancement, CI/CD improvements, customer switching tools, Jira reporting dashboards, security proof-of-concepts, and mobile app development for vehicle management and pet tracking. T-Mobile delivered work up to 3X faster by running parallel work streams. The model required one-third the number of traditional FTEs to deliver the same volume of output. The partnership resulted in 85 projects completed and 56 repeat engagements, demonstrating sustained adoption across multiple initiatives.

Skills

Project Details