At Noesis.tech, I build production-grade generative AI applications for enterprise clients. I engineered a RAG pipeline using LlamaParse, Weaviate, and Gemini that cut agent hallucinations by 75–85%.
I designed agentic workflows for intent classification, tool orchestration, and structured output validation. I also built a multi-LLM gateway for OpenAI, Gemini, Amazon Bedrock, and Groq.
For voice interactions, I engineered a bidirectional Voice AI pipeline using LLMs, Plivo, and Azure Speech, reducing manual sales-agent workload by 70–75%. Its mid-call function calling processed more than 2,000 interactions to extract customer details and classify leads.
I’ve also built omnichannel sales and email workflows, analytics dashboards, and a credit-based billing system. My work connects AI capabilities with customer workflows across web, WhatsApp, Instagram, email, and voice.
