Srijan Singh

I'm building high-scale systems at Samsara, where I reduced primary database load by 30% and eliminated 22-minute tail latencies on 44M-row tables.

I extracted a production gRPC microservice from a monolithic path, routing 6% of production traffic through LaunchDarkly feature flags. I also built privacy-preserving AI usage analytics and reduced average AI token costs by 15%.

I've contributed to JIRA2PR, an autonomous ticket-to-PR pipeline that generated 20% of total merged pull requests from Jira requirements. My work spans microservices, databases, AI workflows, and agentic software engineering.

Previously at Samsung India, I designed an edge U-Net for noise attenuation with 96.4% speech-processing accuracy and used quantization-aware training to deliver a 300% performance boost with less than 0.2% accuracy trade-off. I also build AI automation, documentation agents, and real-time EEG-controlled hardware projects.