Digitally transforming resource allocation for a data annotation leader

Managing a workforce of 5,000 professionals with diverse skills and availability across numerous projects, without delaying onboarding or delivery, was becoming a bottleneck - talent and skill gaps were hard to see, hiring decisions lagged, and information silos slowed HR decision-making. Building the Talent Management System meant integrating it with the client's existing people-management and upskilling systems without disrupting operations, enabling real-time data processing across modules, and adapting it to the client's own workflows and culture.

Zemoso’s discovery pod collaborated with the stakeholders for a week-long GV-inspired design sprint, effectively established the golden user path, and created a high-fidelity, clickable prototype in a two-week sprint.

• Architecture sprint: We adopted the Domain-Driven Design (DDD) approach to identify and prioritize essential entities, services, and Non-Functional Requirements (NFRs). We crafted a hybrid architecture, microservices + event-driven architecture to better meet client’s needs.
   Microservices architecture: For management of RESTful web service. Event-driven architecture: For asynchronous capturing and processing of events occurring in different systems.

• User story mapping: We created user story maps to understand user needs and requirements, and used the t-shirt sizing approach to estimate the effort, define the scope, and prioritize features accordingly.

• Algorithm development: Zemoso built an ML algorithm to ensure a systematic and structured process for resource allocation. The algorithm maximized workforce efficiency through:
  - Skill information management: We used PostgreSQL as the database management system to regularly store and update employees' skill information, leveraging the structured nature of PostgreSQL for ensuring data integrity and reliability.
  - Employee availability management: Managed and tracked employees' availability status in real-time with Node.js’s asynchronous event-driven architecture to fast-track resource allocation.
  - Project requirements and matching: The TMS captured incoming project requirements with a user-friendly React interface and processed all project opportunities with a Node.js back-end, leveraging Kafka event streaming to communicate with all other people management systems.
  - Skills matching and calculation: The ML algorithm utilized the stored skill information and employed computational logic to compare project requirements and skill sets of available employees to provide a clear compatibility assessment.
  - Data-driven decision-making: The TMS displayed the calculated match percentage on the front-end to the Project Managers and enabled data-driven decision making. API Development
  - API development: We developed APIs using Node.js to connect the TMS with the client’s other people solutions. These APIs facilitated seamless communication among various platforms establishing interconnectivity and data flow.

• Robust CI/CD: Throughout the partnership, we implemented CI/CD pipelines to ensure shorter feedback loops and to steadily make progress on product development milestones. Our agile methodologies and continuous feedback loops allowed us to iterate, streamline development, and refine the product.

Zemoso's design and engineering pods built the TMS using a blend of a modern tech stack, architectural strategies, and iterative development methodologies. The client is now testing the Talent Management System internally to assess its effectiveness.