iAllo

Remote

• Built real-time voice AI that runs on live phone calls: streaming Deepgram transcription, intent detection on each utterance, and automated decisions made mid-call
• Designed a LangGraph agent that processes every call transcript with GPT-4o, extracts entities and business insights, and checks its own output before saving results
• Shipped a speech-to-text layer that switches between Deepgram, Google Cloud Speech, and NVIDIA Riva, picking the right provider per call and falling back when one fails
• Wired the telephony stack end to end: SIP calls through Jambonz, Twilio for WebRTC and SMS, dual-channel recording, and live transcripts pushed to the UI over Socket.io
• Cut manual reporting time by 70% with GPT-4 pipelines that return clean structured output and log token usage on every model call
• Rebuilt the billing system with React, TypeScript, and Node.js: Stripe checkout, webhook processing, and idempotency guards to stop duplicate charges

LARRI - Louisville Automation & Robotics Research Institute

Louisville, KY

• Trained custom YOLO-NAS object detection models for a high-throughput industrial sorting facility, tuning hyperparameters and optimizing inference for real-time production deployment
• Designed data augmentation pipelines with motion blur, lighting variations, and geometric transforms to improve model accuracy across changing warehouse conditions
• Built a multi-object tracking system using IoU-based matching and flicker detection to maintain consistent box identification across video frames
• Developed model evaluation tools that compute mAP, precision, and recall on labeled datasets to guide model selection and tuning decisions
• Architected an end-to-end ROS 2 computer vision pipeline that integrates camera streams, object detection, and barcode decoding into a single deployable system
• Containerized ML environments with Docker and set up CI/CD pipelines with automated testing to standardize workflows and speed up team onboarding

ASCC Logistics - Global Enterprise

Nairobi HQ, 20+ International Offices

• Designed and shipped scalable API architectures powering enterprise logistics applications across 20+ international offices
• Developed AI-driven automation systems that improved logistics efficiency by 20% and reduced processing time by 30%
• Led cross-functional deployment of cloud-based AI solutions on AWS and Google Cloud, serving global operations
• Architected high-performance database structures optimized for real-time logistics tracking and analytics