SAURABH SINGH

Hosting Platform for Anthropic Models

• Designed scalable hosting infrastructure managing 20+ models across 50K GPU and 100K CPU instances
• Led 5 engineers to build automated CI/CD pipeline, reducing model validation cycles from 2 weeks to 1 day (95% improvement)
• Implemented automated AMI patching with CVE monitoring, reducing manual effort from 2-3 weeks to 2 days
• Architected version management system and self-healing mechanisms for high availability

FineTuning Platform for Anthropic Models

• Architected and developed LLM fine-tuning platform for Claude models generating ~$20M annual revenue
• Built high-availability distributed systems using ECS, Lambda, Step Functions, and DynamoDB
• Implemented automated testing and monitoring achieving 99.9% platform reliability
• Mentored 4 junior engineers on distributed systems design and operational excellence

Emergency Contact Calling

• Architected voice-activated emergency contact system serving millions of Alexa users via "Alexa, call my emergency contact"
• Designed highly redundant distributed system with multi-region failover for critical situation reliability
• Collaborated with Product, UX, and Legal teams to define technical specifications; implemented comprehensive automated testing frameworks and monitoring solutions

Emergency Helpline Calling (Guard Plus)

• Built mission-critical emergency response platform connecting customers with professional monitoring agents through "Alexa, help me" subscription service
• Designed fault-tolerant architecture with automatic failover handling secure address management, real-time agent dispatch, and third-party emergency service integration
• Reduced connection time by 40% through monitoring-driven optimizations; ensured regulatory compliance across multiple jurisdictions

Roadside Assistance

• Developed voice-activated roadside assistance with automatic location sharing via "Alexa, call roadside assistance"
• Architected highly available integration layer with redundant pathways between Alexa services, location services, and third-party providers using event-driven architecture
• Designed fault-tolerant system with automatic failover and comprehensive error handling to ensure service continuity during critical situations

Audible Content License Microservice

• Redesigned Content License Microservice eliminating single point of failure and enabling independent development and deployment cycles
• Architected fault-tolerant system with improved service resilience and debugging capabilities
• Developed RESTful APIs for Audible iOS application, collaborating with mobile engineers to optimize API contracts and payload structures; implemented monitoring for real-time service health tracking

Integration Platform for Hospital Management Systems

• Architected and delivered SDK-based integration platform (https://developers.practo.com/sdk/java) serving 1,000+ hospitals and 80K+ doctors across South East Asia
• Designed RESTful API architecture with comprehensive documentation; collaborated with hospital IT teams and Product managers to expand platform adoption

Online Appointment Booking and Search

• Developed high-performance backend application serving 100K+ daily user requests; optimized queries and caching reducing API response time by 15%
• Designed scalable doctor search functionality with intelligent filtering and ranking; implemented automated testing and monitoring for peak traffic reliability

Data Processing Pipeline for Healthcare Providers Enlisting

• Built automated data processing pipeline with access control, profile management, and notification services, reducing manual operational effort by 50%
• Architected centralized microservices for OTP verification, secure file upload, and authentication; designed event-driven architecture for real-time notifications with retry mechanisms

Topic Modeling Application

• Developed topic modeling application to identify the most frequent reasons for flight accidents on NASA aviation data using unsupervised Machine Learning algorithms