Hrishikesh “Hrishi” Bhat
I design clearer decisions and operating systems, especially where important judgment is still hidden inside people and processes.
An illustrative diagnostic, not a client case study
A proposed solution becomes useful after we examine the work beneath it.
Automation is a proposed solution, not yet a diagnosis.
Automation is the last step. Understanding is the first.
How I contribute
I make evidence, assumptions, uncertainty, incentives, and trade-offs visible so disagreement becomes something we can examine.
At Sherlock, rule design and escalation paths made validity, severity, and duplicate judgments open to challenge.
A defensible decision has visible reasons.
I turn recurring coordination and judgment into clear ownership, rules, workflows, exception paths, and review boundaries.
Agent-led researcher matching and AI-native bug-bounty operations redesigned the work around the tool.
A tool becomes useful when the work around it has been redesigned.
I explain the structure behind an answer so people can use, question, and improve it independently.
Eight years teaching physics, Python, and Java at Sattva Academy made transfer the test of understanding.
A useful explanation travels without the teacher.
Ideas tested in practice
Situation
I founded and ran Sattva Academy for eight years, teaching physics, Python, and Java while carrying curriculum and institutional operations.
Hidden system
The real test was not whether a learner could repeat an answer. It was whether they could recognize the same structure when the surface changed.
What I changed
I made the reasoning visible, connected lessons to practice, and designed the surrounding schedule and curriculum to support independent use.
What it taught me
An explanation is complete when someone can use, question, and adapt the model without depending on the teacher.
Situation
Across 250+ security engagements, including more than 100 audit contests at Sherlock, decisions carried technical consequences with money and trust at stake.
Hidden system
Validity, severity, and duplicate judgments depended on criteria that had to remain credible when researchers and clients disagreed.
What I changed
I designed rules, judging practices, and escalation paths that put the standard and the reasons for a decision in the open.
What it taught me
Trust under disagreement comes from inspectable reasoning, not artificial certainty or confidence in a single reviewer.
Situation
At Spearbit and Cantina, I worked on agent-led researcher matching and AI-native bug-bounty operations across intake, triage, coordination, and payout.
Hidden system
Generating text was not the bottleneck. Ownership, context, handoffs, exceptions, and the boundary for human judgment governed the work.
What I changed
I designed agents around coherent responsibilities, enabling researcher matching to complete in a couple of hours and routine coordination to move with less friction.
What it taught me
AI becomes useful when the workflow changes and ambiguous decisions still have an explicit human-review path.
A working lens
What do we actually know, and what are we merely assuming?
Where is judgment happening without being named?
What should technology carry, and what should remain human?
Can another person use the model without me?
Beyond the immediate problem
Physics taught me to look for underlying models. Teaching taught me to transfer them. Moving through software, security, and AI systems taught me to enter unfamiliar disciplines without pretending to know them too soon.
For sixteen years, meditation has been an operating practice. Yoga and philosophy deepen the same habit: notice assumption, stay with uncertainty, and protect the attention difficult work requires.
I care about systems because they shape what people must hold in their heads. A good system carries repeatable coordination while leaving meaningful judgment with the person who has the context to use it.
Find the model
Make it transferable
Build the mechanism
Make judgment defensible
Redesign the work
Bring me something difficult
Tell me what is happening, who it affects, and what you have tried. You do not need a polished brief.