All projects
Infrastructure In production

Video platform: encoding, VOD and IPTV

~290 TB of HLS on three origin servers, GPU encoding and live channels

My role
Author · DevOps
Period
2022 — now

The video infrastructure of SPlay and the e-learning platforms: an FFmpeg + NVENC encoding pipeline, HLS library storage and delivery on three mirrored origin servers, TV channel ingest over multicast and SRT with GPU transcoding, broadcast recording.

What I did

  • ~290 TB of HLS VOD on three servers with mirror copies
  • GPU encoding: several video and audio qualities, preview sprites, AES-128
  • IPTV origin: multicast/SRT → GPU → HLS and a relay to a cable operator
  • Channel recording in 15-minute segments
  • HLS caching on the edge proxies

Stack

  • Linux
  • Nginx
  • FFmpeg
  • NVENC
  • HLS
  • SRT
  • MediaMTX

More in this category

Infrastructure Internal system

Company-wide monitoring platform

Metrics, logs, alerts and a status page for ~85 servers, ~70 projects and ~190 services

  • Sole author: 158 of 158 commits
  • ~85 servers, ~70 projects, ~190 services and ~320 domains with DNS and TLS checks
  • Hardware monitoring: IPMI, SMART, RAID, fans, GPUs
  • A /deep-health/ standard with a Django reference implementation, rolled out to a dozen projects

Author · DevOps 2026

  • Python
  • Django
  • Docker
  • VictoriaMetrics
  • Grafana
  • VictoriaLogs
  • Alertmanager
  • +1
Details
Infrastructure Internal system

GPU servers for LLMs

8× NVIDIA B200 and 8× RTX PRO 6000 Blackwell: language models, image and video generation, speech

  • Model evaluation (GLM, Qwen, DeepSeek, Kimi, MiniMax) and benchmarks
  • An LLM gateway with strict routing, virtual keys and budgets
  • Incident runbooks: driver reloads, CUDA graph crashes, a warranty GPU tray replacement
  • OS installs through the BMC and virtual media — no physical access needed

DevOps / MLOps 2026

  • Linux
  • Docker
  • NVIDIA GPU
  • SGLang
  • vLLM
  • LLM gateway
  • ComfyUI
  • +1
Details