Sahil B. Lukhi

Led development of CallIntel - Real-Time Conversational AI Platform with persistent memory, RAG-based knowledge retrieval, and tool-triggered workflows (CRM, messaging, DBs), plus enterprise BYOK support. Architected multi-tenant orchestration with org-level concurrency, prioritization, retries, and fault-tolerant distributed workers via Docker, Kubernetes, and CI/CD. Implemented a sub-second voice streaming pipeline (Twilio/Plivo/Elision) with bidirectional streaming, audio format conversion, VAD, and async buffering to achieve ~700 ms end-to-end latency and natural turn-taking. Built a pluggable AI interaction layer combining multiple model providers and custom STT/TTS pipelines with Supabase persistence for context-aware conversations, transcript storage, analytics, and automation hooks. Developed KnowledgeFlow AI for document ingestion with semantic chunking, hot-swappable embeddings, and 10K+ pages/hour indexing. Designed a hybrid retrieval architecture across dense embeddings and

Performed statistical analysis and historical market studies; applied machine learning techniques to forecast market movements and support strategy optimization. Automated data preprocessing workflows and developed predictive models to assist trading decisions.