Future AI Product Architect

About the company

Clevai is a leading provider of GenAI learning products for working adults and online Math and English tutoring for K-12 students in Vietnam. Clevai is expanding globally with its GenAI products and aims to become a top-5 online education platform in Southeast Asia within five years.

• Clevai received investment from Altara Ventures, FEBE Ventures, and FJ Labs and was recognized as one of the "Emerging Giants in Asia-Pacific 2022" by HSBC and KPMG.
• Clevai became a Top-3 online math provider just two years after its founding by combining celebrity Dual-Teacher livestreaming, small-group Adaptive Tutoring, and vibrant gamification to deliver courses rated 4.8/5.
• Clevai is a pioneer in applying technology to education and the organizer of the TechForStudy seminar series, featuring over 20 speakers, including technology leaders from VinAI, Earable, and Ohmnibot, as well as scholars from Harvard and Oxford. Several of these experts serve as company advisors.
• Clevai's founding team previously built a platform with 2 million paid users and 3,000 online courses across three countries.

Your tasks

3. AI-NATIVE INFRASTRUCTURE Leverage Clevai's fully AI-native engineering infrastructure:

• Custom-built Software Engineering Process Orchestrator: Coordinate and optimize AI-native software development workflows.
• 50-Agent/Skill Fleet: Cover the complete AI-native development lifecycle, with continuous upgrades through compounding loops and team contributions.
• Cloud Knowledge Base: Provide team-based permissions, a comprehensive technical semantic layer, and integrated business context.
• Cross-Agent Communication: Enable real-time coordination through cloud-based agent messaging and shared intelligence.

Level 1: AI Builder (Approximately 6–12 Weeks, 4–6 Projects) Human in the Loop: One Person Ships a Squad's Sprint

• Independently deliver the scope of a traditional development squad's Sprint.
• Apply Karpathy-style agentic engineering workflows.
• Master the AI-native development cycle:
• Specifications (Specs)
• Plan Review
• Code Review
• Evaluations (Evals)
• Build a strong foundation in AI-assisted engineering, product delivery, and quality assurance.

Level 2: AI Lead Builder (Approximately 10–30 Weeks, 5–10 × 2 Projects) Human ON the Loop: One Person Delivers Epics and Initiatives

• Independently deliver complex Epics and Initiatives across multiple development streams.
• Apply agentic engineering at scale using Peter Steinberger-style parallel agents.
• Leverage harness engineering and compounding engineering methodologies.
• Orchestrate multiple AI agents to independently design, build, test, and deliver large software systems.

Level 3: AI Product Architect (After 12–32 Weeks) Owner of the Outer Loop: One Person Runs a User Segment's Full Lifecycle

• Take end-to-end ownership of an entire user segment's product lifecycle, including:
• AI-driven engineering and product development
• Product rollout and launch
• User awareness and acquisition
• User engagement and retention
• Outside-app activities
• Performance measurement and analytics
• User feedback loops
• Continuous product upgrades
• Be evaluated based on user and business outcome metrics rather than the number of features or products shipped.
• Apply advanced industry practices, including:
• LinkedIn-style Full Stack Builder
• A16Z-style Growth Engineering
• Anthropic-style Forward-Deployed Engineering

A. Some Completed Projects 1. AI Agent Calibration and Persona Fleets

• Batch calibration of 20 functional agents using 20 Persona Fleets built from social media crawling.
• Delivery: 2 weeks, 1 person.

2. System Architecture Decomposition and UI Optimization

• Decomposition of a 60-screen system into four architectural layers and a reusable component library.
• Enabled rapid A/B testing, layout customization, and design system personalization for different user segments.
• Delivery: 2 weeks, 1 person.

3. Full-Lifecycle Product Generation Upgrade

• Built and upgraded a complete product lifecycle, including product development, launching, communication, user engagement, analytics, and daily improvement cycles.
• Delivery: 4 weeks, 1.5 persons.

B. Some Upcoming Projects 1. Human-in-the-Loop Signal Pipeline Optimization

• Triage and convert 15 Human-in-the-Loop signal pipelines to continuously calibrate 20 functional agents.
• Estimated delivery: 2–4 weeks, 1 person.

2. Large-Scale Data Architecture and Performance Optimization

• Optimize data schemas and pre-launch system performance.
• Implement transaction integrity and telemetry layers for a 150-table system.
• Estimated delivery: 2–4 weeks, 2 persons.

3. Fine-Tuning Open-Weight AI Models

• Fine-tune open-weight models using private human feedback data.
• Transform sparse feedback signals into versioned training datasets.
• Develop held-out parity evaluations.
• Evaluate the economic break-even point between self-hosted models and API-based solutions.

4. AI-Agent-Driven Experimentation Platform

• Build an agent-operated experimentation platform capable of generating and screening over 100 UX variants per day.
• Use simulated-user fleets to evaluate and rank UX experiments.
• Promote winning variants to live bandit tests under predefined guardrail metrics.

Requirements

A. Common Requirements

• Strong first-principles analytical thinking, with the ability to independently design original system architectures and technical solutions rather than blindly following industry conventions.
• Ability to rapidly Zoom In and Zoom Out (ZIZO) when analyzing complex problems.
• Strong judgment in balancing long-term strategic objectives with short-term execution priorities.
• Experience at a US technology company, a Southeast Asian unicorn, or a high-growth technology company is a plus.
• Education from a world-class engineering university is an advantage.

B. Starting as AI Builder

• Strong Product Owner with some coding knowledge, or a strong Software Developer with product management capabilities.
• At least 5 years of relevant professional experience.
• Ability to bridge technical execution and product thinking to deliver meaningful business outcomes.

C. Starting as AI Lead Builder

• Fulfill all AI Builder requirements.
• Experience managing teams of 10 or more people.
• Proven experience shipping complex end-to-end software products.
• Hands-on experience with AI-native development tools such as:
• Claude Code
• OpenAI Codex
• Google Antigravity

Skills:

• Software Installation
• Software Documentation
• Software Quality Assurance
• Software Development
• Software Testing
• Software Engineering
• Software Architecture
• Teamwork
• ERP Software
• Postman Software

What we offer

Build, Launch, Manage and Iterate Software Products with a 100% AI-Native Engineering Compounding Loop

• Full-time position combined with a structured talent development program through project rotations.
• Learn on the job to independently deliver a squad's Sprints, multiple squads' Epics and Initiatives, or complete user-lifecycle projects across multiple product and business teams.
• Develop exciting B2C and B2B applications designed to serve millions of users.

7. BENEFITS

• Competitive compensation with performance bonuses every six months.
• Exceptional career and compensation growth opportunities.
• Salary increase speed and stock options for top-performing talents (Top 10%) are six times those of regular employees (Lower 35%).
• Stock options are available to 90% of mid-level managers and experts and 20% of lower-level employees.
• A distinctive Silicon Valley-style working culture featuring:
• Radical Candor: Open, direct, and constructive communication.
• "Voi" Structuring: Structured thinking and problem-solving.
• Data-driven OKRs: Quarterly objective setting and performance measurement.
• "Stone Your Boss" Round Tables: Open feedback sessions every six months.
• "Ice Tea": Direct conversations to resolve workplace conflicts.
• Monthly All-Hands Meetings: Company-wide alignment and transparency.