On-chain investigation and evidence systems
I work on crypto-linked matters where chain activity must become usable evidence. The task is not to produce a convincing story from a dashboard. It is to establish what the records show, preserve how they were verified, separate fact from inference, and state clearly what remains unresolved.
Work for decisions that carry consequences
Crypto-linked investigations often move between technical records and professional decisions. Counsel, compliance teams, investigators, exchanges, and leadership need more than a transaction graph. They need a clear account of the relevant records, the reasoning applied to them, the evidence that supports each material point, and the limits that remain.
My work combines on-chain tracing, evidence design, AI governance, and practical communication. The objective is a defensible next step, not an overconfident conclusion.
What makes an investigation hold up
The workflow begins by defining the question, the decision it must support, and the role of the person requesting the work. It then preserves the underlying material, verifies material chain claims against authoritative records, and keeps each stage distinct.
- Record: establish the transaction, contract, address, or other underlying fact.
- Interpret: distinguish a supported analytical inference from the record itself.
- Test: keep competing hypotheses visible and check whether the evidence discriminates between them.
- Limit: stop before unsupported claims about identity, control, recovery, or criminality.
Evidence provenance, verification status, and confidence matter because a technically plausible answer is not enough when another professional must review or rely on it.
BlockIntel Systems: the evidence layer
BlockIntel Systems is a private intelligence and evidence system for deterministic chain data, tracing, screening, and case support. It helps turn dispersed public-chain records into a bounded, reviewable evidence workflow without confusing data retrieval with professional judgment.
Its role is to make the work repeatable: source-bounded facts, transaction and address context, verification records, and a clear path from material evidence to a stated finding or open question. It does not make attribution determinations by itself, and an AI output is never treated as an autonomous conclusion.
Keshav OS: controlled AI-assisted work
Keshav OS is the private operating environment around the work. Systems are isolated by purpose, source-of-truth records govern their state, and verification rules shape handoffs between steps. This makes the workflow inspectable without turning a private investigation environment into a public product demonstration.
The principle is straightforward: computation can be deterministic, AI can assist research and review, and the investigator remains accountable for the conclusion. That separation protects both the quality of the work and the people who rely on it.
Tools that make the method visible
Some systems are public because they support learning, monitoring, or transparent technical practice. They are related to the work, but they are not substitutes for a case-specific investigation.
Helix Signal is an open-source, self-hostable stablecoin and chain intelligence platform. KeySense is a source-available, client-side workbench for understanding multi-chain key derivation and address paths. Both make their boundaries clear.
Training and translation
I translate tracing and evidence work into practical sessions for investigators, lawyers, compliance practitioners, and forensic teams. The aim is not to create more dashboards. It is to help people ask better questions, understand what a chain record can prove, and know when additional legal, exchange, or investigative process is required.
This work includes invited crypto-tracing and evidence sessions for SVPNPA Hyderabad and Mumbai Police, alongside c4 Academy work.
What is deliberately not public
Private source code, client material, system access, provider configuration, infrastructure details, and case-specific operational records are not published. The public material demonstrates the method, the systems, and the standard of reasoning without exposing work that must remain controlled.