Systems & AI Engineer · Law Student
Durgapur, West Bengal, India · X (@anisayakmitra) · Instagram
I build local-first autonomous AI runtimes and developer tooling in Rust and Python, with a focus on deterministic execution, process sandboxing, and verifiable authority boundaries. Alongside software engineering, I study law and explore the intersection of algorithmic governance and system security.
- AGENT-PANDORA — A local-first autonomous AI agent platform written in Rust.
- Runtime Architecture: Modular workspace spanning 21 discrete runtime engines for planning, memory retention, sandboxing, and verification.
- Governed Execution: Reference-monitored authority chain enforcing one-shot effect permits, fail-closed audit ledgers, and process confinement.
- Orchestration & Interfaces: Durable multi-agent job queues, isolated Git worktree subagents, interactive Ratatui TUI, WebAssembly extension packages, and stdio MCP integration.
Active contributor across systems, AI infrastructure, and developer tooling ecosystems:
- AI & Agent Infrastructure: BerriAI/litellm, awslabs/stickler, Nano-Collective/nanocoder, Podiom/Podiom, reticlehq/reticle
- Systems, Rust & CLI Tooling: servo/rust-smallvec, vyncint/termlens, mabd-dev/reposcan, yottadynamics/yottacode
- Data & Workflow Pipelines: apache/airflow, Hebbian-Robotics/hflow, Open-audit-foundation/Open-Audit
- Languages: Rust, Python, Go, TypeScript, SQL, Shell
- Systems & Runtime: Tokio, WebAssembly (Wasm), Linux Sandboxing, Ratatui (TUI), SQLite (WAL), Docker, Git
- AI & Protocols: Model Context Protocol (MCP), Local & Cloud LLM Orchestration, Deterministic Evaluation, Structured Output Validation


