Devanshi Desai

About

Full-stack/product-leaning engineer in logistics and high-traffic portals who ships production AI features: built an AI-assisted shipment status Q&A system using Pinecone + GPT-4 and a high-volume Python ingestion pipeline (500K+ records/day), delivering 35% fewer support tickets and cutting resolution time from 11 to 4 minutes. Also led a legacy Angular-to-React/TypeScript rebuild that boosted Lighthouse performance from 60 to 90, and has hands-on AWS EKS operations experience including resolving a 3x traffic scaling incident.

Experience

Education

Key Strengths

• Shipped AI-assisted shipment query resolution system (Pinecone + GPT-4) reducing support tickets 35% and resolution time 11→4 minutes
• Built reliable high-volume ingestion service processing 500K+ records/day with <0.2% error rate
• Production AWS EKS + Docker + CI/CD (GitHub Actions/CodeDeploy) with 99.9% uptime at high transaction volumes
• Improved retrieval quality and latency via embedding strategy, metadata filtering, hybrid search, reranking, and prompt constraints to reduce hallucinations
• Led modernization from legacy Angular monolith to React + TypeScript component library; Lighthouse performance 60→90
• Strong UX engineering: robust loading/error/empty states, request cancellation for race conditions, debouncing, infinite scroll, micro-interactions
• Designed performance-focused PostgreSQL schema for e-commerce catalog with indexing strategy validated via EXPLAIN ANALYZE
• Handled real production scaling incident (3x traffic spike) by diagnosing CPU/pod pressure and scaling Kubernetes node groups; added scheduled scaling and improved alerting