Open to opportunities

Paweł Izdebski

@paweizdebski

Message

Computer Vision architect building production AI & agentic systems for real-time video, MLOps, and scalable streaming.

Poland

Message

What I'm looking for

I’m looking to design and ship production computer-vision and agentic AI systems—architecture-first for real-time video—partnering with teams that value scalable engineering, measurable performance, and robust MLOps (LangGraph/RAG/streaming pipelines).

I build AI systems that run in production—not on slides. My foundation is Computer Vision, trained through an MSc in Optoelectronics & Informatics Systems, then compounded by 15+ years of real infrastructure experience at CANAL+, where I scaled streaming systems and led engineering teams.

Computer Vision isn’t just a skill I picked up—it’s the discipline I was trained in. On the AI side, I ship agentic architectures like LangGraph into production, with an architecture-first mindset and practical tooling across LLM development and observability.

I’ve led end-to-end delivery for high-stakes, real-time media systems: a production sports-telemetry pipeline using YOLOv9 + ByteTrack; a DAI Manifest Manipulator (HLS/DASH/SCTE-35) delivering sub-50 ms TTFB under CDN load; and CV-based watermarking and content fingerprinting for forensic traceability and piracy detection.

I also bring engineering depth to delivery constraints and adoption, from edge optimization (YOLO quantized via ONNX to TensorRT for sub-30 ms inference) to lecturing at the University of Warsaw. My professional ethos is simple: build the architecture, instrument the pipeline, and make performance and reliability measurable.

Experience

Work history, roles, and key accomplishments

Current

Guest Lecturer (AI & Data)

Current

University of Warsaw

Feb 2026 - Present (4 months)

Designed and delivered a graduate course covering LLM architectures, RAG systems, MLOps pipelines, and computer vision applied to media.

Retrieval Augmented Generation (RAG)MLOps Computer Vision For Media

Current

Head of Product Innovation

Current

Spyrosoft

Mar 2025 - Present (1 year 3 months)

Architected and led an AI-native transformation for a 200+ person media-tech organization by migrating workflows to LangGraph-based agent-orchestrated pipelines, reducing manual overhead across engineering.

LangGraph Multi agent orchestration MCP Computer Vision YOLOv9

Current

R&D Expert (Edge Computer Vision)

Current

3PSOFT

Optimized CV models for constrained edge hardware by quantizing YOLO variants and exporting via ONNX to TensorRT, achieving sub-30 ms inference on embedded GPU devices without cloud dependency.

YOLO Model Quantization ONNX Runtime TensorRT Object Detection On Device Event Recognition

Solution Architect (AI & Ad-Tech)

GPW (Warsaw Stock Exchange)

Mar 2024 - Aug 2025 (1 year 5 months)

Engineered an ad-insertion DAI manifest manipulator (HLS/DASH/SCTE-35) achieving sub-50 ms TTFB under full CDN load, supporting live ad insertion at scale.

HLS Dash SCTE 35 SSAI DAI Amazon SageMaker MLFlow LangGraph Multi Agent Systems

Computer Vision / Media Architect

Adspective

Jan 2024 - Oct 2024 (9 months)

Architected the core CV pipeline for virtual product placement, using CLIP-based frame scene analysis and brand-fit scoring to enable photorealistic product insertion at scale.

Virtual Product Placement Brand Fit Photorealistic Insertion

Computer Vision / Media Architect

Vestigit

Jan 2023 - Oct 2024 (1 year 9 months)

Pioneered a real-time CV-based video watermarking system enabling perceptually transparent embedding and forensic extraction for distribution-chain source tracing across broadcast and OTT.

Watermarking Computer Vision ML Content Fingerprinting Broadcast Security OTT Security

Head of DevOps & CDN

Canal+ Poland

Jan 2020 - May 2022 (2 years 4 months)

Managed a 50+ person engineering organization across DevOps and software development, owning technology strategy, delivery governance, and vendor orchestration.

DevOps OTT Streaming Infrastructure Architecture Live VOD Delivery Team Leadership

Broadcast Network Architect

ITI Neovision / CANAL+ Cyfrowy

Dec 2012 - Dec 2019 (7 years)

Designed and operated nationwide DTH and IPTV infrastructure serving millions of subscribers, progressing from broadcast network architect to CDN and IT solutions manager over 7 years.

Broadcast Engineering DTH Infrastructure IPTV Delivery CDN Network Architecture Systems Operations

Education

Degrees, certifications, and relevant coursework

Military University of Technology (WAT)

Master of Science (MSc), Optoelectronics & Informatics Systems

Completed an MSc in Optoelectronics & Informatics Systems at the Military University of Technology (WAT), with an emphasis on image processing as a hard engineering discipline.