Responsibilities
- Develop and enhance a React-based web application across key product areas such as authentication, session setup, workspace navigation, file review, agent activity streaming, results analysis, and user administration.
- Design reusable component libraries for complex, state-dependent workflows that ensure maintainability, accessibility, and ease of extension for development teams.
- Build and support backend API endpoints using Express and TypeScript to manage session coordination, file handling, workspace operations, usage metrics, and new product features.
- Connect frontend components with backend services for authentication, LLM routing, usage tracking, agent orchestration, cloud storage, and PostgreSQL data management.
- Transform intricate backend and agent states into intuitive UI patterns, including handling empty, loading, error, review, and resumable workflow states.
- Enhance real-time communication between backend, agent runtime, and user interface using server-sent events and event-driven architectures.
- Display incremental agent outputs including token-by-token text generation, tool execution summaries, plans, task lists, progress indicators, cost tracking, file modifications, warnings, and final outcomes.
- Manage the lifecycle of streaming connections, including retries, cancellation, cooperative stopping, resumption, error recovery, and clear user feedback during long-running agent processes.
- Define event specifications that allow the UI to clearly represent agent behavior while abstracting away implementation complexity from end users.
- Design user experiences and API structures that clarify agent actions, file changes, review-ready outputs, and tasks requiring human judgment.
- Lead improvements in the custom multi-turn agent loop involving message transmission to model providers, streaming response parsing, tool execution, result observation, and iteration within isolated cloud environments.
- Build proprietary tools to extend agent functionality in areas such as file operations, data analysis, transformation, visualization, document generation, validation, and workflow automation.
- Expand the containerized runtime to support additional programming languages, libraries, utilities, file formats, analytical techniques, and integrations required by expert users.
- Define clear tool schemas, permission controls, workspace access rules, input validation, and error messaging to ensure safe and effective agent tool usage.
- Create scalable patterns for integrating new tools without compromising runtime stability, transparency, or debuggability.
- Support integration across multiple AI models including OpenAI, Anthropic, and other frontier or local providers, accounting for differences in message formats, tool calling, streaming, structured outputs, and error handling.
- Develop translation layers that standardize interactions across model providers while preserving access to each model’s advanced capabilities.
- Maintain prompt and context systems that guide agent behavior, including analytical identity, methodology adherence, interaction modes, tool policies, quality benchmarks, and escalation procedures.
- Implement features for token estimation, usage monitoring, context compression, conversation summarization, prompt caching, and model selection in extended analytical sessions.
- Analyze how agent behavior varies across models, prompts, tools, and workflows, then refine the system to improve quality, cost-efficiency, speed, reliability, privacy, and user trust.
- Own the end-user lifecycle of AI work sessions from creation and configuration through file upload, execution, interruption, resumption, result review, and cleanup.
- Build browser-to-cloud file transfer systems supporting multi-file uploads, progress tracking, validation, browsing, previewing, downloading, and handling of large or diverse file types.
- Develop interfaces and APIs that help users track remote workspace states, generated outputs, intermediate artifacts, source files, and final deliverables.
- Strengthen the link between workspace and agent states to improve user clarity and engineering debuggability.
- Implement resilience strategies including retry mechanisms, rate limit handling, tool error recovery, graceful cancellation, resumable workflows, and failure reporting.
Benefits
- Comprehensive total rewards program featuring a strong benefits package
- Wellness initiatives supporting physical, mental, emotional, and financial health
- Internal immigration support for international employees and business travelers
- 100 hours of annual training through formal and informal learning programs
- Career mentoring and performance coaching from a designated senior colleague
- Opportunities for leadership and collaboration through internal development activities
Work Arrangement
Hybrid
Other
- Employees are expected to work in the office for at least 3 to 4 days per week, which may include travel to other company offices or client sites.
- The company offers a comprehensive total rewards program including a strong benefits package, wellness support, and internal immigration assistance for international staff.
- Annual training allowance of 100 hours is provided through a mix of formal courses, informal learning, technical training, presentation skills, seminars, mentoring, and coaching.
