Overview
You're joining as one of the first engineers at a 6-person startup building AI data governance infrastructure. The product scans enterprise knowledge bases (documentation, wikis, support content) to find contradictions, outdated info, and data quality issues that break AI systems. You'll be writing the code that enterprises like Disney depend on to safely deploy AI. Small team, big customers, lots of greenfield work.
Role Snapshot
| Aspect | Details |
|---|---|
| Role Type | Founding/early engineer (full-stack or backend-focused) |
| Product Stage | Early production - have paying customers but still building core features |
| Tech Complexity | High - NLP, data parsing, enterprise integrations, AI/ML infrastructure |
| Team Size | You + founder(s) + 1-2 other engineers |
| Customer Impact | Direct - you'll talk to Disney's team, debug their issues, ship features they requested |
| Scope | Broad - database design, API development, data pipelines, customer integrations |
Company Context
Stage: Seed (just raised from Susa Ventures and Wischoff Ventures)
Size: 6 employees - mostly engineering
Growth: Landed Disney + other enterprises, now scaling to handle demand
Market Position: First mover in AI-specific data governance - defining the category
What You'll Actually Build
Core Product Areas
- Data Ingestion: Connect to customer knowledge bases (Confluence, Notion, SharePoint, custom wikis). Parse different formats, handle auth, incremental updates.
- Quality Detection: NLP/ML to identify contradictions, outdated content, missing information, ambiguous language that confuses AI systems.
- Remediation Workflows: Tools for data teams to review issues, approve fixes, track changes over time.
- Governance Layer: Audit trails, version control, approval chains for regulated industries.
- AI Integration Layer: APIs that AI tools query to verify data quality before using information.
Day-to-Day Reality
- Feature Development (50%): Ship new capabilities customers are asking for. Disney needs X integration, another customer needs Y report.
- Customer Support (20%): Debug why ingestion failed for a customer's data source. Their Confluence has weird formatting. Figure it out.
- Infrastructure (15%): Scale pipelines to handle larger datasets. Optimize queries that are timing out. Set up monitoring.
- Customer Calls (10%): Join technical discussions with enterprise data teams. Explain how the system works, what's possible, what limitations exist.
- Architecture Decisions (5%): Weekly syncs with founders on product direction, tech choices, what to build next.
The Honest Reality
What's Hard
- Enterprise data is messy: Every customer's knowledge base is organized differently, uses different tools, has different edge cases. You'll spend a lot of time handling one-off scenarios.
- Quality is subjective: "Contradictory information" is easy to say, hard to define algorithmically. You'll iterate on detection logic constantly based on false positives.
- Customer timelines: Disney wants a feature by their deadline. You're balancing shipping fast for customers vs building the right foundation.
- Small team reality: When something breaks in production at 8pm, you're probably the one fixing it. No one else knows that part of the codebase.
- Ambiguous requirements: Customers say "we need better governance" but can't articulate exactly what that means. You're translating vague asks into concrete features.
What Success Looks Like
- Ship features that directly enable customer renewals and expansions
- Build integrations that unlock new enterprise logos ("we'll buy if you connect to our X system")
- Reduce production incidents and customer support load through better reliability
- Make architectural decisions that scale as dataset size and customer count grows
Tech You'll Work With
Likely Stack (infer from problem space):
- Backend: Python (ML/NLP libraries), possibly Go for performance-critical parts
- Data: PostgreSQL or similar, vector databases for semantic search, data pipeline tools
- ML/AI: NLP models for text analysis, LLM integration for quality detection
- Integrations: REST APIs for enterprise tools (Confluence, SharePoint, etc.)
- Infrastructure: Cloud (AWS/GCP), Docker/Kubernetes, CI/CD
What You'll Learn:
- Enterprise data governance patterns
- NLP/ML in production at scale
- Building reliable data pipelines
- Selling/explaining technical products to non-technical buyers (you'll be on sales calls)
Who You'll Work With
Internally:
- Rebecca Wang (Founder, Stanford CS dropout) - product direction, customer relationships
- Akylai Kasymkulova (Co-founder) - likely technical co-founder, architecture decisions
- 1-2 other engineers - split the codebase with you
- Eventually GTM hires - you'll explain product capabilities, what's feasible to promise
Customers:
- Data engineers at enterprises debugging ingestion issues
- Compliance/legal teams asking how audit trails work
- AI/ML teams integrating your APIs into their systems
Requirements
- 2-5 years software engineering experience, preferably backend or data engineering
- Strong Python skills (likely primary language for ML/NLP work)
- Experience with data pipelines, APIs, or enterprise integrations
- Comfortable working directly with customers (explaining technical concepts, debugging their issues)
- Okay with ambiguity - you're building something new, not following established patterns
- Interested in AI/ML applications (don't need to be an ML expert, but should want to learn)
- SF-based or willing to relocate (6-person team needs in-person collaboration)