Vadim
Vashkelis
I help companies turn AI, computer vision, and video research into systems that work in production.
λx. research→productionλx. research(x) → production(x)The Backstory
I lead R&D teams that turn AI, computer vision, and video algorithms into production systems — from research prototypes to deployed infrastructure.
Over the last 15+ years, I've worked across video intelligence, autonomous systems, large-scale AI pipelines, hardware-aware inference, and engineering leadership.
My work combines research depth with production constraints: data, latency, cost, reliability, deployment, and teams that actually have to ship.
Advising companies on AI infrastructure, production ML, and how not to let complexity outrun common sense.
Embedded Intelligence Lab
Applied research advancing practical AI — computer vision, compact language models, inference optimization, multimodal systems, and hardware-aware deployment for resource-constrained environments.
What I Work On
Production AI Systems
Turning models, pipelines, and prototypes into reliable systems that can run at scale.
Computer Vision & Video Intelligence
Detection, tracking, video analytics, compression, and visual understanding for real-world environments.
Edge & Embedded AI
Optimizing AI for constrained hardware, accelerators, and deployment outside the cloud.
AI Strategy & Technical Leadership
Helping teams choose the right architecture, roadmap, infrastructure, and delivery model.
SOME THINGS I'VE BUILT
Automation Through Conversation
Led research and engineering for voice AI agents and automation systems that translate business communication — phone and email — into dependable end-to-end workflows processing thousands of interactions monthly.
Visual Intelligence at Scale
Led large-scale image and video analytics platforms processing terabytes of daily data, with training and inference pipelines running on 100+ GPUs for high-throughput production search across millions of indexed images.
Intelligent Video Compression
Led video processing R&D for intelligent video storage systems handling petabyte-scale archives, combining AI-accelerated analytics, codec optimization, and deployment on specialized hardware.
Computer Vision Center
Founded and led a computer vision center growing to 20+ specialists, building internal expertise across machine learning, image processing, data pipelines, and applied research delivery.
Large-Scale Video Platforms
Owned and evolved multimedia software stacks serving millions of devices across server-side and device-side systems in a nationwide production environment.
Autonomous Driving Systems
Founded and led an 8-year R&D company producing multiple patents in scene understanding, 3D analysis, and robotics-oriented perception for autonomous systems.
Selected Publications
N. Trukhina, V. Vashkelis
N. Trukhina, V. Vashkelis
A. Popov, N. Trukhina, V. Vashkelis
V. Vashkelis, N. Trukhina
N. Trukhina, V. Vashkelis
N. Andreasyan, M. Struve, A. Popov, M. Nikolaev, V. Vashkelis
V. Vashkelis, N. Trukhina, S. Kumar
V.V. Vashkelis, M.V. Korman
V. Vashkelis, M. Korman
Tools & Experiments
vLLM Wizard
A small tool for estimating and configuring vLLM deployments across GPU types and model sizes.
AI Compression Notes
Experiments on compressing long AI conversations using semantic prompts and LLMs as decoders.
TTS Model Comparison
Side-by-side comparison of text-to-speech models for quality, latency, and production readiness.