Velum Labs

The OS for data quality across any stack

Winter 2026ActiveB2BMachine LearningData EngineeringSan Francisco, CA, USA

Company

Velum is the operating system for data quality. We observe real data traffic to automatically derive contracts, trace lineage across any stack, and enforce integrity from ingestion to executive dashboards - no manual rules, no migration required. Data trust you can prove. From the pipeline to the boardroom.

Verdict

High Signal

Market Opportunity

Data quality/observability is a well-established B2B enterprise market with TAM north of $5B. The ICP is clear — data platform teams at mid-to-large companies with multi-team analytics stacks (dbt, Snowflake, BigQuery). Semantic drift and inconsistent metric definitions are a real, documented pain point that Airbnb, Uber, and others have spent years solving internally. Monetization path via SaaS tooling for data teams is well-proven.

Medium Signal

Founder Signal

Benjamin Muñoz-Cerro attended Stanford (2021-2025) and Harvard for physics — essentially a fresh grad with strong academic pedigree and research affiliations (CERN competition, Samsung Solve for Tomorrow winner) but no industry work experience visible. CTO Alen Rubilar-Muñoz has no LinkedIn available, described only as 'ml swe | hackathon winner' in YC bio. Academic credentials are impressive but neither founder appears to have shipped production software at a real company or sold to enterprise data teams.

Medium Signal

Competition

Direct competitors include Monte Carlo, Soda, Great Expectations, Atlan, and dbt's own semantic layer (MetricFlow). The traffic-observation-first approach (no manual rules) is a meaningful differentiation vs. rule-based tools like Great Expectations or Soda. However, dbt Labs, Databricks (Unity Catalog), and Atlan are well-funded and moving fast in semantic layer territory. The 'PR-native enforcement' angle is novel but easily replicable by incumbents.

Low Signal

Product

No named customers, no pricing page, no live demo, no revenue or usage metrics. CTA is 'Become a design partner' — they are still in pre-product/design partner stage. The product 'Alma' is described conceptually with a clear technical vision, but there is zero evidence it exists in production or has been deployed anywhere.

OverallC Tier

Velum Labs is attacking a real, high-value problem in data quality and semantic consistency, and the technical approach — learning contracts from query traffic rather than manual definitions — is genuinely differentiated. However, the product appears to be pre-revenue vaporware at this stage: no customers, no live product, design partner CTA only. Both founders are fresh academic grads with strong research backgrounds but no evident track record of shipping enterprise software or selling to data teams. The market is crowded with well-funded incumbents (Monte Carlo raised $236M, dbt has MetricFlow). The team needs to convert design partners to paying customers fast, or this risks being an academic project that never finds PMF.

Active Founders

Benjamin Muñoz-Cerro

Founder

CEO @Velum Labs | stanford quantum computing | harvard physics

Alen Rubilar-Muñoz

Founder

CTO @ Velum Labs | ml swe | aspiring mathematician | hackathon winner | world citizen

Velum Labs

TierC Tier

BatchWinter 2026

Team Size2

StatusActive

LocationSan Francisco, CA, USA