Harry Dai

Backend Software Engineer

Experience

TikTok · Backend Software Engineer Intern

San Jose, CA

  • Consolidated livestream quality evaluation services into a Go platform driven by RocketMQ. Pluggable stages let business teams assemble workflows in a web UI without code changes.
  • Instrumented throughput, latency, and error rates in Prometheus. Built Grafana dashboards and configured alerts for volume shifts, infrastructure issues, and end-to-end latency.
  • Traced a spike in scoring latency using service logs, queue lag metrics, and CPU flame graphs. Resolved it by migrating overloaded pods.
  • Built a ClickHouse-backed dashboard that lets teams analyze real-time quality scores by dimensions such as CDN vendor, region, and video tier.
  • Added audio scoring with an FFmpeg-based service that detects abnormal silence in segments fetched from S3.

TikTok · iOS Software Engineer

Chengdu, China

  • Fixed unexpected multi-device call disconnections by matching signals to device IDs. Related complaints fell from five per month to zero.
  • Led an RTC optimization that adjusted video resolution to call window size, reducing downstream bandwidth costs by about $50K per month. Coordinated the rollout with Android and server teams.
  • Refactored four core messaging features into reusable components and shipped them in an SDK adopted by three other apps.

TikTok · iOS Software Engineer Intern

Chengdu, China

  • Eliminated photo-album flicker by loading low-resolution previews first and requesting full-resolution media only for viewing or editing.
  • Shipped chat improvements that increased chat daily active users by 600K, including background customization and drag-and-drop media sharing.

Project

LiveBridge RTMP

  • Built a livestream server that relays audio and video from publishers such as OBS to players such as VLC.
  • Implemented RTMP handshakes, stateful chunk parsing and message reassembly, and AMF0 commands for connection, publishing, and playback.
  • Enabled viewers to join live streams by replaying cached metadata and codec headers, then waiting for the next video keyframe. Verified H.264 and AAC playback with FFmpeg.

Education

Northeastern University

M.S. in Artificial Intelligence · Boston, MA

Nanjing University

B.S. in Computer Science and Technology · Nanjing, China

Technical skills

Languages
C++, Go, Java, Python, Objective-C, TypeScript
Backend and cloud
Microservices, REST APIs, MySQL, Redis, Kubernetes, S3, Kafka, RocketMQ
Observability
Prometheus, Grafana, metrics instrumentation, alerting, CPU profiling