Name: 高级爬虫工程师 / Senior Web Scraping Engineer
Price: 108 EUR
Availability: InStock

Languages

English
Native or bilingual
Chinese
Native or bilingual

Workplace preferences

Remote only

Primarily works remotely

Company
Senior Python Engineer (Protocol Reverse Engineering)
TRANSPORTATION
May 2024 - Today (2 years and 1 month)
Shangqiu, China
Led the design and delivery of an end-to-end automation engine for the RCS (Rich Communication Services) protocol. By
reverse-engineering Google Messages' gRPC stack, the system handles bulk account registration, activation, renewal,
and end-to-end encrypted messaging at scale.

Key contributions:
1. Reverse-engineered 9 core gRPC endpoints of the Google Messages client (checkin, register, getConfiguration,
getContent, setContent, sync, proceed, ACS, restore). Hand-wrote the Protobuf schemas and built the full request
pipeline, achieving a stable end-to-end registration flow.
2. Designed a Frida 16.x injection layer for the GMS process. A FridaGateway singleton manages per-device sessions and
hooks DroidGuard tokens, registration keys, and contentSig signatures at runtime, solving the core problem that
anti-abuse tokens cannot be forged offline.
3. Built a Redis-based scheduler (task queue, SOCKS5 proxy pool, engine control channel) paired with a MongoDB state
machine, decoupling the registration and renewal pipelines and making them recoverable and observable.
4. Built a multi-worker async engine on asyncio + httpx (HTTP/2) governed by per-device token slots (3 concurrent
slots per device), increasing single-host daily account throughput several-fold.
5. Implemented an E2EE subsystem on top of the Signal Protocol: X3DH key agreement, HKDF derivation, AES-256-CBC
encryption, plus a server-streaming gRPC long connection consuming IMDN receipts; covering the full send / receive /
ack pipeline for encrypted messages.
6. Built an ADB device-cluster ops layer (5-minute watchdog + Streamlit dashboard) and a per-worker logging system
using ContextVars that auto-injects worker_id / device_id, significantly cutting triage time under heavy concurrency.

Tech stack: Python 3.13, asyncio, Protobuf / gRPC, httpx, Frida, MongoDB, MySQL, Redis, Streamlit, Signal Protocol.
自己
Senior Web Scraping Engineer / 高级爬虫工程师
February 2026 - May 2026 (3 months)
Shangqiu, China
【English】

Owned reverse-engineering and anti-bot countermeasures for an automated carrier-subscription system handling hundreds
of thousands of daily orders.

- Protocol reversal: Analyzed minified webpack chunks and official SDKs to recover method-parameter contracts; built
an SDK-injection pattern that bypasses partner front-ends and cuts cold-start by 3-4x.
- Self-healing decryption: Decoded the SDK's enc field (HMAC + DES/AES) with multi-regex key extraction and a two-tier
cache; auto-detects weekly obfuscation rotations.
- Fingerprint impersonation: Used Go uTLS to match Chrome's JA3 + HTTP/2 frame ordering, defeating server-side TLS
detection; serialized queues + warm-up reloads resolved concurrent-session collisions.
- Risk-signal recovery: Cross-referenced captures, logs, and DevTools to reverse 26 risk rules — backend timestamp
cross-checks, isTrusted gestures, Bezier trajectories — driving key error rates to zero.
- Engineering: Codified reverse-engineering know-how into a resumable, auditable onboarding workflow, shrinking
new-channel integration from days to hours.

Stack: Node.js, Egg.js, Playwright, Go (uTLS), MongoDB, Redis.

【中文】

负责咪咕运营商订购系统的逆向接入与反风控对抗，日均订单数十万。

- 协议逆向：分析压缩混淆的 webpack chunk 与官方 SDK，还原核心方法入参契约，落地 SDK 注入模式绕过业务方前端，冷启动提速
3-4 倍。
- 加密自愈：对咪咕 SDK enc 字段（HMAC + DES/AES）做多正则 key 提取 + 双层缓存，自动感知每周 SDK 混淆更新。
- 指纹对抗：用 Go uTLS 模拟 Chrome JA3 + HTTP/2 帧序列，解决服务端 TLS 指纹检测；串行队列 + 预热 reload 解决并发
session 抢占。
- 风控信号还原：交叉对比抓包 / 日志 / DevTools，反推服务端时间戳交叉校验、isTrusted gesture、贝塞尔轨迹等 26
项风控规则，把核心错误码命中率压到 0。
- 工程化沉淀：把逆向经验固化为可中断 / 可恢复 / 可追溯的接入工作流，新渠道由数天压到小时级。

技术栈：Node.js / Egg.js / Playwright / Go (uTLS) / MongoDB / Redis