English Textbooks, Reimaginedby AI Agents
LingoLift (Story Factory) × LingoForest (School OS) — an AI education agent that turns English textbooks into readable, listenable, chatable stories and closes the loop on how students learn. A 4-person team scaling to 8–10 in the next 3 months.
One Closed Loop
A content factory and a school OS, wired together into a single learning ecosystem.
LingoLift — Story Factory
Feed an English textbook to AI agents and get a complete story experience out: graded outlines → chapter stories → stylized illustrations → consistent characters → CosyVoice voiceover → WeChat review gate.
LingoForest — School OS
After students read, Elo/Ability scoring and CAT placement kick in. The model recommends the next book, teachers see learning analytics, and an AI companion chats with students.
Our Technology Stack
An all-Go backend with a ByteDance Eino agent layer on top of Alibaba models.
Not Starting From Zero
The architecture is built and running at a working prototype. Here's what already works.
8 Stable SFE Workflows
Outlines, chapters, quizzes, illustrations, voiceovers, characters, and the full pipeline — all producing reliably.
Verified Output Quality
3,000-word chapters with illustration quality reviewed and accepted as better than raw LLM output.
4-Identity RBAC
Role-based routing for user / admin / teacher / editor / student is already cut.
Learning Analytics Triple
Session Elo, attitude_score, ability_score, and level_code, with Redis NX 90s sliding-window rate limiting.
3 Hard Constraints Written Into Spec
Unified L-Levels
One grading scale across the whole product.
LLM Never Grades
AI produces content; humans and models with verifiable rules score students.
SFE Approved → SGS Publish
Nothing ships to students without passing the content factory gate.
What's Left
This is why this hiring page exists. The hard engineering is still ahead.
SFE Polishing
- Dual-mode chapter merge (batch / per-chapter pause)
- Character visual consistency across chapters
- Skill registry
- WeChat review gate
SGS Progression
- P3 student insights
- P4 full learning analytics
- P5 platform: 5s group aggregation, TeacherAsk
Frontend Gaps
- Student AI companion chat tab not yet wired
- Teacher group insights still templated
- SGS P4–P5 frontend sync
Production & Compliance
- Single-node to cluster (MySQL / MongoDB / Redis)
- Monitoring and alerting
- WeChat ecosystem integration
Hiring Priorities
First an AI Agent engineer and two Go backends to push SFE/SGS to P4. Then two frontend engineers to align all three clients. DevOps and QA to finish.
AI Agent Engineer (Senior)
Key blocker: Character consistency, dual-mode chapter merge, LLM audit
Backend Engineer (SFE, Go)
Key blocker: 35 Tools consolidation, Redis Stream, WeChat review gate
Backend Engineer (SGS, Go)
Key blocker: RBAC, learning models, SFE ↔ SGS sync
Frontend Engineer (Teacher/Editor)
Key blocker: Group insights, Ask, RBAC, SGS P4–P5 sync
Frontend Engineer (Student)
Key blocker: Library, reader, AI companion chat tab
DevOps / SRE
Key blocker: MySQL/Mongo/Redis clusters, monitoring, AI inference
Benefits & Culture
A small team that builds things that matter — with the freedom to work the way you want.
Remote Friendly
1–2 monthly meetups in Guangzhou (Panyu / Tianhe), primarily remote
Equity Options
Every role gets equity (0.1%–0.5% early-stage, by role and start date)
13-Month Salary
Competitive base with year-end bonus tied to company performance and role KPI
AI-First Workflow
Encouraged to use AI tools — Cursor, Claude, our own products
No Timesheets
No time tracking, no weekly reports; async collaboration via Linear + Feishu
Learning Budget
¥5,000 per person per year for technical / product learning
Flexible Holidays
10 days annual leave on joining, plus extra Spring Festival leave
Product Pride
Not ordinary SaaS — an AI content factory producing craftsmanship
Hiring Process
Four steps, offer within 48 hours of the final interview.
Initial Screening
HR or founder reviews resume + GitHub / portfolio, 30-minute conversation
Technical Interviews
2 rounds, 60 minutes each — first on tech stack fundamentals, second a project deep-dive
CTO + Founder Interview
30-minute conversation about product vision and cultural fit
Offer
Offer extended within 48 hours
A Closed Loop Nobody Has Walked Yet
Five codebases, three learning models, and eight workflows are already built. Come run the loop with us.