Looking for a job

Laido Unknown

@laidounknown

Message

AI engineer at UpMedia automating video production with evaluated LLM agents for 28 channels.

Kazakhstan

Message

What I'm looking for

I’m looking for full-time work as a B2B contractor or an employee via Deel EOR, with remote work and full EU-hours overlap.

At UpMedia, I built and run agentic LLM systems for a YouTube media company with 28 channels in five languages. My automated pipeline handles about 200 videos a week, and evaluation runs on a golden set before prompt or model changes ship.

I built RAG code search for AI agents across a seven-service platform, improving recall@5 from 0.53 to 0.70 on held-out questions. I also rebuilt the script writer as a structured, multi-step pipeline; verbatim-duplicate frames fell to zero and fact collection became 60 times cheaper.

At Dorodnicyn Computing Centre, Russian Academy of Sciences, I built a pupil and iris boundary detector and fine-tuned YOLO on synthetic data for QR codes and barcodes. I’m a PhD candidate in Computer Science at MIPT, with a thesis on object detection in 3D images.

Experience

Work history, roles, and key accomplishments

Current

AI Engineer

Current

UpMedia

Jan 2024 - Present (2 years 9 months)

Sole engineer building agentic LLM pipelines for a YouTube media company, handling about 200 videos weekly. Developed evaluation gates, RAG code search, and multi-step script writer, reducing costs and errors.

Current

Research Engineer, Computer Vision

Current

Dorodnicyn Computing Centre, Russian Academy of Sciences

Jan 2021 - Present (5 years 9 months)

Built a pupil and iris boundary detector using U-Net and IrisDenseNet, achieving IoU 0.972 at 62 ms per image. Fine-tuned YOLO on synthetic data for QR codes and barcodes, achieving 94% recall with one false alarm in 4,000 images.

Computer Vision U Net IrisDenseNet YOLO PyTorch Synthetic Data IOU Object Detection