Personal Research & Engineering Archive

Tanishq Dasari

I am a Computer Engineering student and backend engineer from Solapur, Maharashtra, India. I enjoy building systems that remain understandable as they grow.

Most of my work revolves around backend infrastructure, static analysis, AI governance, runtime policy enforcement, developer tooling, and knowledge systems. Rather than building isolated applications, I design infrastructure that helps developers build safer and more reliable software.

Selected Impact

  • Built Anchor: AST compliance scanning and isolation engine supporting 118 governance rules.
  • Published Zenodo Preprints: Authored academic and implementation briefs detailing deterministic execution boundaries.
  • Regulatory Contributions: Submitted formal policy feedback and briefs to the FSB and RBI.
  • Apache-2.0 & MIT: Open-source projects and packages published on PyPI (anchor-audit, shadowwatch).

Primary focus

AI governance, static analysis, runtime policy enforcement.

Key projects

Anchor, Canon, AnchorGrid, and Shadow Watch.

Verifiability

Every claim is backed by open-source code and papers.

Work

Flagship Projects

I document my systems work as complete engineering case studies rather than simple summaries.

Anchor

Flagship

AI governance engine with static analysis, runtime enforcement, replay, and decision audit chains. Published on PyPI.

Read Case Study →

Canon

Flagship

Governance synchronization system for monitoring frameworks, diffing rule changes, and proposing updates.

Read Case Study →

Governance Hub

Flagship

Operational platform for governance domains, organizations, verification, and centralized audit chains.

Read Case Study →

QuantForge

Infrastructure

AI-powered financial terminal with developer-first CLI analytics, instrumented with Shadow Watch telemetry.

Read Case Study →

Invariants

Engineering Principles

Deterministic logic

Software should be understandable before it is clever.

Explicit design

Good architecture reduces complexity rather than hiding it.

Documentation as code

Every design decision should be reproducible, reviewable, and explainable.

Read Principles Page

Journal

Engineering Journal

Lab notebook logs exploring AST patterns, static analysis false positives, and compiler tooling design decisions.

Governance as Code for Autonomous Systems

Why policy must be executable, versioned, and replayable if autonomous systems are expected to behave inside real-world constraints.

Jun 12, 20261 min read
Read essay

Runtime AI Safety and Deterministic Enforcement

A technical note on enforcing policy at runtime without losing determinism, observability, or replay fidelity.

Apr 27, 20261 min read
Read essay

Human-in-the-Loop Governance for Model Risk Management

A note on making governance updates reviewable, contextual, and safe enough for enterprise and policy environments.

Feb 11, 20261 min read
Read essay

Milestones

Timeline

Released Anchor v5

Shipped the latest major version of Anchor with static analysis, runtime enforcement, deterministic auditing, and constitutional enforcement capabilities.

Jul 1, 2026

Released Canon v0.1

Launched the governance synchronization system for monitoring frameworks and generating human-reviewed evidence packages.

May 12, 2026

Submitted RBI technical consultation

Provided technical consultation on AI model risk management and governance controls for autonomous systems.

Mar 8, 2026

Built Governance Hub

Developed the management platform for governance domains, replay systems, audit chains, and enterprise governance operations.

Jan 18, 2026

View Timeline

Contact

Get in touch

I am open to discussions about software architecture, AI risk frameworks, and systems engineering.

View Contact details