TikTok LIVE · Agency Operations Intern · Paris

Hi, I'm Jianghao.

I work through partners, and I turn the data behind them into systems.

Sciences PoMaster's in Marketing

French · DALF C1BA in French, GPA 92, 2nd of 24

English · IELTS 7.5Chinese native

How I read this role

A B2B2C job, where the last C is creators.

01 · Communication

Many stakeholders, one direction

Creators, agencies and TikTok teams all want something different. Someone has to keep them moving together.

02 · Data

Monitored daily

Activity, revenue and health for every agency. Problems show up in the numbers before anyone calls.

I've worked this way three times

10+ travel platforms · Universal 10 e-commerce platforms · L'Oréal Dozens of channel partners · Lark

Universal Beijing Resort · Trade Marketing

A ¥46M co-op budget, read week by week.

11

partner platforms, incl. Ctrip and Meituan

¥46.2M

total co-op budget in 2025

1.7B

impressions and 45M clicks

2.73%

click-through rate, +0.31 pt YoY

Where the impressions movedshare in 2025 · change vs 2024

Zone 4 · core region

48.6%+2.6 pt

Zone 5 · long-haul

20.6%+6.5 pt

Zones 1–3

27.6%−5.8 pt

Nationwide

3.1%−3.2 pt

Spend went deeper into Jiangsu–Zhejiang second-tier cities and long-haul markets.

Students on campus

regional projects were college roadshows

+4%student visitors YoY during the season

12 days a week of manual data cleaning

→

2Automated in Python, with dashboards

→

3Mid-sized platforms dipped before every holiday

→

4Spend moved 1–2 weeks earlier, now team standard

L'Oréal · 3CE · Trade Marketing

Three teams, one table.

Gifts kept running out before big sales. Nobody was slacking. Each team only saw one piece.

Sales✓ demand ✕ stock

Supply chain✓ arrivals ✕ priorities

Warehouse✓ live stock ✕ matched to demand

One SKU-keyed table → Power BI

View 1 · stock vs demand, per channel

View 2 · projected stockout date

Channel C · in 4 days→ transfer stock before it runs out

Channel names and values are placeholders.

70%→95%

Channel gift requests fulfilled. From firefighting to planning ahead.

8,000+

sets sold in week one of the 3CE × Sephora 3.8 campaign. I was part of it from start to finish.

ByteDance · Lark · Global Marketing, APAC

APAC events and content, from plan to pipeline.

Events

~20

APAC events I helped coordinate

4 AWS Summits5 AI Seminars10+ client sessions

Leads

600+

leads per AWS Summit, at the two I ran end to end

Summit 1

Summit 2

each dot ≈ 100 leads

Sign-ups

+50%

AI Seminar registrations vs Q1

indexed, Q1 = 100

Content

+20%

impressions on the Customer Success Cases series, CTR +1 pt

indexed, before = 100

LinkedIn A/B test · same period, one country per postwinning format highlighted

🇻🇳 VietnamPractical AI news & tipsCustomer case studies

🇮🇩 IndonesiaPractical AI news & tipsCustomer case studies

Initiative · built with AI at Lark

Spot the friction. Build the tool.

01 · Lead checker

Partner leads that pass first time

Partners upload leads. An AI agent checks fields and removes duplicates before they reach the CRM.

50%→90%+

first-time submission success

02 · Event radar

Find events before they find us

Scans Southeast Asian B2B events daily, scores each one on four evidence signals, and ranks them.

Invites → active scan

from waiting for invitations to scanning the market, used by my manager and me

03 · Materials generator

Event assets without the queue

Fill in a form, and brand-safe templates from the design team produce banners, posters and invites.

3 days→2 hours

working days to hours, open to the whole team

🏆2nd place · ByteDance hackathon, with an AI project

● LIVE

Putting it together

What this job needs, and what I'd bring.

Communication

Talk with agencies

I've managed partners at Universal, L'Oréal and Lark. I'm comfortable doing it in French and English.

Data

Read their health

Dashboards, root causes, early warnings. I've built all three.

Initiative

Build as it scales

When a process keeps breaking, I turn it into a tool. AI makes that fast.

Thank you. Merci.