Envive AI
Senior Full-Stack Engineer / Senior Software Engineer
Mar 2024 – Nov 2025
ReactTypeScriptNext.jsFastAPILLM Tooling
Impact
15%LCP / TTI improvement
20%P95 render time cut
8%RLHF task time reduction
Design System
- Led design system and UI architecture using React and TypeScript with Storybook.
- Standardized tokens, theming, and accessibility to WCAG 2.2 across the product.
AI Agent Tooling
- Built agent consoles with multi-turn state, tool-use traces, response diffs, and guardrail explanations so brand teams could review and approve changes without engineering support.
- Shipped LLM observability dashboards with trace samples, bad-case triage, and evaluation runs tied to brand KPIs and safety scores.
Performance & Edge Rendering
- Implemented edge-rendered product and search UIs with Next.js and ISR; added prefetching and reduced hydration work, improving LCP and TTI by 15%.
- Owned evaluation UI with run dashboards, error analysis, slice-and-compare views, and version diffing; reduced P95 render time by 20% on analyses with 100k+ rows.
Back-End & APIs
- Added FastAPI and Node routes for catalog, returns, and order APIs with schema validation, rate-limited webhooks, and redaction/policy checks before responses.
- Wrote Python workers for evaluation aggregation and lineage; added signed URLs, dataset ACLs, and retry logic with dead-letter queues for unreliable jobs.
Safety & RLHF
- Built red team scenario builders and judge/reviewer tools with task queues, rater QA surfaces, and integrated safety taxonomies with clear pass/fail specs.
- Streamlined labeling and preference workflows for RLHF with keyboard-first review, conflict resolution, and gold-set audits; reduced per-task time by 8%.
- Unified model evaluation results and SEAL leaderboards into a single compare view for ranking by capability and safety, improving decision speed for release reviews.