AI Full-Stack Software Engineer

Architected and optimized real-time voice AI agents using LiveKit, TTS/STT services (Deepgram, Cartesia) and GPT-4/5 for low-latency interactions. Improved call quality with a 4.8/5 satisfaction rating. Developed real-time audio pipelines managing interruptions, silence detection, and turn-taking. Led integration of LiveKit with telephony systems (Twilio, FreeSWITCH) enhancing call stability and reducing latency. Implemented jitter and packet loss reduction strategies. Utilized LiveKit analytics for debugging and resource management. Built adaptive dialogue pipelines with VAD, ASR → LLM → TTS improving automation accuracy by 40%. Integrated voice AI into healthcare workflows reducing manual effort by 45%. Orchestrated and fine-tuned LLMs leveraging Amazon Bedrock and Bedrock Data Automation.

Led full-stack development for the first AI social network enabling humans and AI characters to coexist through posts, comments, DMs, and interactive experiences. Built core features for realistic character generation using ComfyUI, Stable Diffusion, and custom training pipelines. Designed and implemented LLM-powered AI chatbots, story generation, and prompt management systems. Developed text-to-image generation workflows integrated with social feed. Engineered cross-platform mobile experiences with React Native and iOS development. Architected scalable backend services using Node.js, Supabase, and AWS. Optimized event-driven microservices for scaling, low-latency, and secure multimodal AI outputs. Collaborated on product launches increasing user adoption.

Built generative AI solutions for healthcare operations including automatic email drafting, clinical note summarization, and visual content generation. Designed internal automation systems reducing manual effort by 35%. Developed voice-interactive AI medical agents, receptionists, and telemedicine assistants using ASR and TTS. Created AI receptionist handling inbound calls, verifying patient info, and routing calls. Built telemedicine AI agent for pre-visit info collection and note generation. Integrated AI microservices into production SaaS with secure data handling and role-based access. Exposed AI capabilities as APIs consumed by web apps and internal tools. Collaborated with clinical and operations teams to translate requirements into AI features. Implemented monitoring and experimentation for AI features.

Built AI video analytics features for Cyclops including automated avatar overlays, voice narration syncing, and rapid localization of insights for retail clients. Created script-to-video workflows generating narrative reports from store camera data. Developed real-time preview pipelines for visitor traffic and cloud rendering for BI reports, reducing delivery time by 60%. Implemented secure backend for RTSP camera streams, demographic tracking, and role-based dashboards. Integrated real-time alerts and POS data for staffing and campaign evaluation. Partnered with product teams to scale video-to-insights pipelines supporting 150k+ daily videos.