Vinoo Ganesh

Vinoo Ganesh is the CEO and co-founder of Kepler, which builds verifiable AI for financial services. His approach assigns language models the work of interpreting financial questions and planning analyses while deterministic software retrieves source data, performs calculations, and records how every result was produced.

From Palantir to institutional investing

Ganesh spent seven years at Palantir, moving between core engineering and deployments for defense, intelligence, and commercial customers. He worked on search and indexing infrastructure and the company’s Spark and YARN compute platform, then developed Project Frontline to place software engineers in forward-deployed roles and bring firsthand customer experience into product development.

He subsequently became chief technology officer at Veraset, a geospatial data company that was later acquired. His work migrating data products from CSV to Apache Parquet examined how infrastructure choices affect data quality, delivery costs, and customers’ workflows. At Citadel, he led business engineering for Ashler, overseeing investment platforms, data pipelines, and the teams supporting them.

Ganesh founded Kepler with John McRaven, another former Palantir engineer, to bring that operational experience to institutional finance. Their founding vision focuses on work products such as investment memoranda, discounted cash-flow models, and fairness opinions, where plausible prose is useless if the underlying figures cannot be defended.

• Atomic provenance: Models identify where a financial figure originates; deterministic tools extract and preserve the actual value. Every number remains connected to its underlying filing, internal document, or other authorized source.

• Scope determinism: Language models decide what needs calculating, but databases, parsers, and conventional software retrieve inputs and perform the arithmetic. That separation makes results auditable while avoiding unnecessary spending on model computation.

• Derivation chains: Calculated metrics retain an inspectable history of their inputs, transformations, and institution-specific assumptions. Verification establishes that an answer follows a particular firm’s rules, even when two investors interpret identical evidence differently.

• Forward-deployed engineering as product strategy: Engineers embedded with customers observe actual workflows, ship narrowly scoped solutions, and turn field discoveries into durable product capabilities. Ganesh developed this approach through deployment experience and applies it at Kepler.

His writing on reliable AI context extends this emphasis to lineage, versioning, validation, reproducibility, and observability. Ganesh also teaches data and AI engineering, advises startups, and serves on Washington University’s McKelvey Engineering National Council. He sees the same verification architecture extending into legal research and healthcare, where professionals must trace consequential conclusions back to dependable evidence.