Santekno
Projects/KLS Live Project — Seller Side
Backend / APILive

KLS Live Project — Seller Side

Scaled high-value livestream seller operations with batch registration, validation, and monitoring for major campaigns across the region.

2026-kls-live-seller · main
# E-Commerce / Livestream
$ go -X POST /2026-kls-live-seller/v1/sync
→ 201 Created
{ "status": "ok" }
publish 2026-kls-live-seller.synced
Role
Backend Engineer (Campaign)
Timeline
2026
Team
4
Domain
E-Commerce / Livestream
Stack
10 tech
Status
Live
01

Overview

The KLS Live Project (high-value livestream) was a strategic initiative to grow livestream revenue during major campaigns by improving recruitment, session registration, and operational visibility. I worked as the backend campaign engineer for the seller console side, designing and delivering the backend needed to scale seller participation and go-live readiness across the region.

A backend has no screens, so its "UI" is the operational surface: the registration API, the batch/ops flow, and campaign dashboards.

API contract — session registration & template fields
Ops view — batch upload & job monitoring
Grafana — go-live rate and campaign health
03

Problem & Solution

Problem

The existing process could not comfortably support roughly a 5x increase in eligible sellers, nor lift go-live rates from a low baseline. Registration validation, template-based uploads, batch submission, and go-live tracking all needed stronger backend support to become scalable and observable.

Solution

I assessed the end-to-end seller flow and delivered solutions across four repositories: new HTTP/RPC contracts, additional enumeration fields (Estimated Revenue, Expected Ads Spend, platform username, Session Level), multi-region handling, and a new scheduler workflow for batch session registration. I integrated creator validation with shop-channel-service, supported FE testing, absorbed mid-project requirements, and released the seller side with broad self-test coverage.

04

Workflow

Three actors interact with the service; the diagrams are written in English to stay consistent across both languages.

flowchart diagram

register session

upload template

batch submit

register

validate creator

Seller

Seller Console

Campaign Backend

Scheduler

Budget Management Platform

Shop Channel Service

Use case interaction
sequence diagram
Budget Management PlatformScheduler JobCampaign BackendSeller ConsoleBudget Management PlatformScheduler JobCampaign BackendSeller Consoleupload session Excelvalidate rows + creatorenqueue batch workflowregister sessionsgo-live statusresult summary
Sequence flow
05

Architecture

flowchart diagram

rpc

metrics

Seller Console

Seller Campaign Service

Campaign Core Service

Campaign Registry Service

Campaign Ops Service

Scheduler

Budget Management Platform

Shop Channel Service

Prometheus / Grafana

System architecture
Architecture notes

seller-campaign-service accepts uploads and validates rows and creators (via shop-channel-service), then enqueues batches into the scheduler, which registers sessions on the budget-management platform and reports go-live status. campaign-core-service, campaign-registry-service, and campaign-ops-service handle downstream registration and operations. All flows run across multiple regions with Prometheus/Grafana monitoring.

state diagram

rows + creator ok

batch job

platform accepted

go-live

uploaded

validated

submitted

registered

live

Session registration lifecycle
06

Features

Seller
  • Online session registration & validation
  • Excel template generation & upload
  • Go-live status visibility
Operations
  • Batch session submission via scheduler
  • Campaign monitoring across regions
System
  • Multi-region logic
  • New scheduler workflow for batch registration
  • Creator validation via upstream service
  • New API contracts & enum fields
07

Tech Stack

language
Go
framework
RPC frameworkHertz
scheduler
Scheduling Framework
protocol
RPCHTTP
database
Managed RDS (relational database)
infra
Multi-region
observability
PrometheusGrafana
08

Engineering Challenges

01
Scaling seller participation roughly 5x

Participation had to support roughly a 5x increase in eligible sellers reliably. Fix: a scheduler-based batch submission flow with structured validation and upstream creator checks, keeping throughput high without manual bottlenecks.

02
Mid-development requirement changes

New fields (e.g. platform username) arrived during development. Fix: adaptable template-generation logic and enum design so the Excel pipeline could absorb changes without rework.

03
Consistency across four repositories

Logic spanned `seller-campaign-service`, `campaign-core-service`, `campaign-registry-service`, and `campaign-ops-service`. Fix: clear ownership boundaries plus broad self-testing across all regions and scenarios.

09

Outcomes

5x
target seller participation
4
repositories delivered across
50+
commits shipped
5 wk
delivery duration
10

Lessons Learned

Scaling an operational workflow needs more than new endpoints — it needs careful batch-processing design, clear ownership across repositories, and close alignment with FE, PM, and QA to keep multi-region delivery stable.

11

Related Santekno Tutorials

Interested in working together?

Order, payment, or high-traffic event pipelines — let's talk architecture and trade-offs.