Sweekar M

I’m an AI-focused backend engineer with hands-on experience building LLM-powered applications, multi-agent workflows, and production backend systems using Python. I’ve prioritized production-grade reliability, security, and observability—especially when integrating LLM capabilities into real services.

In my engineering-heavy internship at CyberARC MSP / CarcMail, I refactored a monolithic backend into a feature-first vertical-slice architecture across 10+ domain modules. I optimized authentication, asynchronous workflows, and API performance, then added Prometheus-compatible metrics for requests, latency, errors, and background services.

I build and ship AI experiences through projects like SoulOS, where I implemented multi-model routing, tool execution, and persistent session memory, and an AI Portfolio Assistant with a RAG pipeline for document ingestion and contextual answers. I also create multimodal assistants using real-time transcription, OCR-based screen understanding, and WebSockets for streaming interaction modes—grounded in scalable agentic workflow design.