Mirwise Khan

Offer on-demand consultation on LLM, Agentic AI, Multi-Agent Systems, Video Generation to VCs in Silicon Valley.

Attended Generative AI summer school at Oxford ML, with focus on LLM, Agentic AI and Multi-Agent Systems with speakers from Google DeepMind and Amazon AWS.

“Exploring Image Representations of Time-Series for Enhanced Model Performance”, NEURIPS 2024 [Publication Under Review]. Benchmarked the performance of different image-based time series methods.

Invented optimal frame selection techniques for 3D reconstruction using generative AI (GANs) to generate detailed 3D models (NeRF – Neural Radiance Field) from 2D images 3D in a constrained environment.

Deployed generative AI-based deep fake vision models (PyTorch) on VPS with Docker for an entertainment app with over 10K downloads.

Compared Apache Kafka & Flink to open-source alternatives, proposed a data architecture migration plan saving 20% cost while coordinating with team of 5 data engineers. Used Terraform and Kubernetes (K8) to automate docker container orchestration and server scaling.

Implemented hand gesture recognition using computer vision AI foundation models on Android device, to work in offline setting.