Frontend Engineer, Vision
Bengaluru • FullTime
Posted 27d ago
Remote Work Policy
On-site
Employment Type
FullTime
Categories
Applied AI Engineer
About the job
Sarvam is building India's sovereign AI platform, focusing on research, models, infrastructure, and applications. This role is on the team responsible for building the serving harness for in-house vision-language models, turning them into production-ready document intelligence platforms. The goal is to match or exceed the performance of frontier hosted models at a fraction of the cost, handling the complexities of Indian documents at scale. You will be the frontend owner, working closely with backend and applied AI engineers to build the user interfaces that enable human trust and interaction with AI models.
Responsibilities
- Build the document review experience, including PDF and image rendering, navigation, bounding-box overlays, and source-to-extraction linking.
- Develop human-in-the-loop correction workflows for fast, field-level editing and integrate corrections back into the evaluation loop.
- Create the extraction schema designer for users to define, test, and version structured output.
- Build internal evaluation and observability dashboards for accuracy, latency, cost, and model comparison.
- Manage real-time and long-running state, including streaming partial results, job progress, and optimistic updates.
- Own and optimize frontend performance for handling large documents through techniques like virtualization and lazy loading.
- Collaborate with backend engineers to define and shape API contracts.
- Establish frontend standards for the team, including component architecture, typing, testing, and accessibility.
Requirements
- 4-5 years of experience building production frontend applications with a strong sense of ownership.
- Deep expertise in React, including hooks, rendering behavior, state management, and performance profiling.
- Strong proficiency in TypeScript, utilizing the type system to prevent invalid states.
- Command of browser platform fundamentals: layout, events, memory, and rendering, with comfort in Canvas or SVG.
- Experience with complex, data-dense product surfaces like dashboards, editors, or review tools.
- Solid API integration instincts, including async state, caching, polling, streaming, and error handling.
- Design sensibility to ship clean interfaces from rough directions.
- Familiarity with PDF rendering in the browser (e.g., PDF.js) or annotation/labelling tools.
- Exposure to AI/ML product surfaces, such as human-in-the-loop review or model output inspection.
- Willingness to contribute to Python or Go backend code when necessary.