A research and engineering direction
In some cases, sensitive information can be processed and analysed without every party receiving the original data used in the computation.
Privacy-enhancing computation can support joint calculations across organisations, operations on encrypted data, and verification of specific claims without revealing the secrets behind them.
Pankkih helps assess whether these methods fit a practical need, and designs and runs bounded technical experiments. We do not claim to develop new cryptographic protocols, and we do not promise production readiness for every algorithm. For technical service management context see Expertise; on cryptographic lifecycle also PQC assessment.
Methods
The methods below solve different problems; they are not interchangeable.
- MPC (secure multi-party computation). Several parties can compute a joint result without revealing their original inputs to one another.
- PSI (private set intersection). Parties can learn which elements their datasets have in common without revealing the rest.
- FHE (fully homomorphic encryption). Computation can be performed on encrypted data without the computing party decrypting it.
- ZKP (zero-knowledge proofs). The truth of a claim can be verified without revealing the secret used in the proof.
Practical example
Two companies each hold a customer database. PSI can identify overlapping identifiers without either side handing over its full table. If the parties instead need an agreed metric (for example a count of shared customers or another aggregate) without exchanging raw records, that task belongs to MPC — not PSI.
Applicability and pilots
Typical work covers the following steps:
- Define the computation task and confidentiality requirements
- Identify the parties and the trust model
- Compare methods against the task
- Review existing libraries and tools
- Run a limited PoC on test or synthetic data
- Measure cost, latency and traffic
- Analyse what the output itself reveals
- Recommend whether further work is justified
Technical references
Discuss a privacy-preserving computation pilot. Related: Services, Expertise, Solutions (data protection), PQC assessment.