Leo Marin

Leo Marin is the co-founder and CTO of Fawna. He leads the engineering organization responsible for turning large volumes of unstructured customer conversation data into fast, secure, searchable, and explainable product insight.

His work spans platform architecture, integrations, data systems, search, applied AI, permissions, security, and the technical foundations required to make feedback intelligence dependable at scale.

Background

Leo has spent his career building data-heavy SaaS products, workflow platforms, and integration infrastructure. His experience includes systems that must combine large volumes of external data, maintain strict account boundaries, support complex user permissions, and remain understandable to the people using them.

He became interested in customer feedback intelligence because the technical problem is inseparable from the trust problem. A system may generate a convincing summary, but teams cannot rely on it unless the source evidence, processing history, permissions, and limitations remain clear.

Role at Fawna

Leo owns the technical direction of the Fawna platform. His responsibilities include:

• Defining platform architecture and engineering standards.
• Building reliable data ingestion from support, CRM, call, review, and collaboration tools.
• Designing data models for conversations, accounts, themes, insights, permissions, and workspaces.
• Developing search, retrieval, clustering, classification, and summarization systems.
• Establishing security, privacy, observability, and operational reliability.
• Partnering with product and design to make technically complex workflows feel understandable.
• Growing the engineering team and its development practices.

The technical challenge behind Fawna

Customer conversation data is difficult to work with because it is:

• Unstructured and inconsistent.
• Distributed across many external systems.
• Rich in context that can be lost during normalization.
• Potentially sensitive and permission-dependent.
• Continuously changing as new conversations arrive.
• Open to ambiguity when automated systems group or summarize it.

Leo’s work focuses on building a platform that creates useful structure without hiding those realities.

“Automation earns trust when people can inspect the evidence, understand the system’s limits, and correct the result when context was missed.”

Engineering principles

Leo guides Fawna’s engineering work through several core principles:

• Traceability: every important insight should remain connected to its source conversations.
• Secure defaults: account boundaries and permissions should be enforced at every layer.
• Observable systems: ingestion, processing, and model behavior must be measurable and debuggable.
• Human control: users should be able to review, refine, merge, split, or reject automated outputs.
• Pragmatic architecture: complexity should be introduced only when it solves a real product or reliability need.
• Graceful failure: external integrations and AI services should fail visibly and recover predictably.

Applied AI at Fawna

Leo approaches AI as one part of a wider product system. Models may help classify conversations, group related problems, generate summaries, detect changes, and retrieve supporting evidence. But the surrounding product must provide:

• Structured inputs and reliable metadata.
• Clear evaluation criteria.
• Versioned prompts and processing logic.
• Quality monitoring and representative test sets.
• Review states and user correction workflows.
• Fallbacks when confidence or coverage is low.

This makes the output more useful and reduces the risk of presenting uncertain automation as fact.

How Leo works across the team

Leo collaborates closely with product design, product insights, customer success, and leadership. Technical decisions are evaluated against user trust, workflow clarity, and long-term maintainability—not only implementation speed.

He encourages engineers to understand the full customer workflow, ask why a capability matters, and contribute to the product decision rather than receiving narrowly defined implementation tasks.

Current focus areas

• Improving multi-source ingestion reliability and data quality.
• Making search and evidence retrieval faster across large workspaces.
• Strengthening theme explainability and review workflows.
• Expanding enterprise permissions, auditability, and data controls.
• Evaluating model quality across different feedback sources and customer segments.
• Reducing operational complexity as the platform scales.

Perspective on trustworthy infrastructure

Leo believes the most important engineering work is often invisible. Customers should not have to think about retries, synchronization, indexing, access boundaries, or processing failures. They should experience a system that is predictable, fast, and honest about what it knows.

His goal at Fawna is to build infrastructure that allows teams to use AI-assisted customer insight with the same confidence they expect from other core business systems.