Vantage Labs is an AI-first incubator with deep expertise in Natural Language Understanding (NLU), powering educational products through over 40 patents in language processing and cognitive computing. During my 8-month tenure as a Senior Software Engineer, I contributed to two major initiatives: enhancing the AI-driven essay scoring platform with a new speech scoring feature, and developing a licensing dashboard for educational institutions using Vantage tools.
Goals
- Extend Vantage’s existing essay scoring system to support spoken language evaluation
- Build a scalable admin dashboard for institutions to manage student licensing
- Improve UX and accessibility of various client-facing experiences
- Maintain engineering discipline while delivering features in a fast-moving Agile team
My Contributions
1. Speech Scoring Integration
Vantage’s essay scoring system was already widely adopted by schools. A new requirement emerged to score students’ spoken responses in different assessment environments.
- Integrated Amazon Transcribe to convert audio recordings into text
- Used Amazon S3 to store and manage user-submitted speech files
- Connected transcribed content to Vantage’s internal essay scoring engine
- Delivered a seamless backend pipeline using Java and Spring Boot
- Enabled the “Speech Scoring” feature within writeSHIFT, enhancing its real-world usability for teachers and institutions
2. Licensing Dashboard for Institutions
I developed a robust admin dashboard to help school authorities manage software licensing for students. The dashboard allowed institutions to:
- Bulk order licenses for Vantage products
- Track license usage, including whether each student activated their license
- View and update license validity periods
Technical highlights:
- Bootstrapped with JHipster, combining Spring Boot and React.js
- Used PostgreSQL as the data store
- Integrated Keycloak for secure authentication and role-based access
- Built UI using Material UI, ensuring accessibility and consistency
3. Browser Extension UI Enhancements
I also contributed to improving customer onboarding by implementing the welcome and marketing screens for the CorrectEnglish browser extension, helping users understand the value of the tool quickly and intuitively.
Outcome
- Extended the platform to support speech-based assessments, increasing product value and versatility
- Delivered a production-ready licensing dashboard, empowering institutions to manage students at scale
- Helped improve first-time user experience for a browser-based product used by thousands
- Maintained high standards of code quality and team collaboration within an Agile setup
Key Takeaways
- Leveraging AWS tools like Transcribe and S3 enabled fast, scalable voice-to-text workflows
- JHipster proved valuable for accelerating admin dashboard development in a secure, maintainable stack
- Close collaboration with cross-functional teams ensured technical solutions aligned with product and business goals
- Speech scoring introduced multi-modal AI capability to a traditionally text-based assessment platform
Tools & Technologies
- Java, Spring Boot, JHipster
- React, Material UI
- PostgreSQL
- Keycloak
- Amazon Transcribe, Amazon S3
- Jira for Agile workflow (Scrum + Kanban)