Skip to content
@akanjilal-work

akanjilal-work

Anirban Kanjilal — work & demos

Frontier compute, made secure, cost governed, and production ready.

Solutions architect with more than twenty years building highly transactional, resilient, and secure systems for regulated industry, on the cloud. This is the open source work behind akanjilal.dev: taking AI and quantum out of the demo and into systems with SLAs, auditors, threat models, and a budget.

Live projects

🛡️ llm-security-lab — OWASP LLM risks, vulnerable vs hardened, mapped to MITRE ATLAS and the NIST framework. Runs live in your browser.

⚛️ quantum-on-braket — runnable, cost aware quantum algorithms on Amazon Braket, free on the local simulator with a spending guard before any real hardware.

💰 ai-finops-agent — an agent that reads FOCUS billing data and surfaces idle accelerators and spend spikes. Runs live in your browser.

🚦 llm-cost-guardrails — per tenant budgets, rate limits, and a spend kill switch for language model apps.

🔌 mcp-secure-gateway — a hardened Model Context Protocol gateway with auth, a default deny allow list, and audit logging.

🧪 agent-eval-harness — unit tests for agent behaviour, gating a deploy on tool, safety, and cost checks.

On the roadmap: agentic-patterns · reference-architectures

📝 Writing and live demos at akanjilal.dev · 🔗 LinkedIn

Popular repositories Loading

  1. llm-security-lab llm-security-lab Public

    OWASP LLM Top 10 and agentic attacks with vulnerable vs hardened variants, mapped to MITRE ATLAS and NIST AI RMF

    Python

  2. quantum-on-braket quantum-on-braket Public

    Runnable, cost-aware quantum algorithms on Amazon Braket (Grover, QAOA, VQE, QPE)

    Python

  3. ai-finops-agent ai-finops-agent Public

    An agent that reads FOCUS-format billing data and surfaces AI and cloud spend anomalies

    Python

  4. llm-cost-guardrails llm-cost-guardrails Public

    Middleware enforcing token and cost budgets, rate limits, and per-tenant cost attribution for LLM apps

    Python

  5. mcp-secure-gateway mcp-secure-gateway Public

    A hardened Model Context Protocol gateway with auth, tool allow-listing, output validation, and audit logging

    Python

  6. agentic-patterns agentic-patterns Public

    Minimal runnable agent patterns (single, supervisor/graph, handoff) across LangGraph, CrewAI, and the Claude Agent SDK

Repositories

Showing 10 of 12 repositories

People

This organization has no public members. You must be a member to see who’s a part of this organization.

Top languages

Loading…

Most used topics

Loading…