Jignesh (Jiggy) Kakkad, Principal AI Engineer in Sydney

I build production AI with security engineered from the start.

I have 21 years of experience across telecommunications, banking and enterprise AI. I build LLM, RAG and agentic systems on Azure and AWS, and bring security engineering into the architecture, APIs, delivery pipelines and operations that support them. My career includes Quantium, x15ventures, Commonwealth Bank and Telstra.

View experience Start a conversation
Experience
21 years across AI, telecommunications and banking
AI engineering
Production LLM, RAG and agentic platforms
Security engineering
Secure SDLC, DevSecOps, API security and AI security

Engineering outcomes

Selected impact

5 times

Faster AI queries

Improved the architecture behind AskTelstra to make production AI queries five times faster.

30+

Services migrated

Moved more than 30 services to a new API gateway with no customer downtime.

40%

Shorter delivery cycles

Reduced deployment cycle time by introducing stronger DevSecOps practices and automation.

1.5M

Transactions each day

Led engineering for services supporting more than 1.5 million daily transactions.

AI and security

AI engineering and security engineering belong together

AI engineering

I build production LLM, RAG and agentic platforms across orchestration, retrieval, evaluation, APIs, integrations and observability. At Telstra, architecture improvements made AI queries five times faster.

Security engineering

I integrate security into design and delivery through secure SDLC, DevSecOps, API security, identity and data boundaries, application testing and production monitoring. I apply the same discipline to risks specific to AI.

Engineering approach

How I engineer secure AI

  1. 01

    Set clear boundaries

    Give agents the minimum tools, permissions and data required for the task.

  2. 02

    Protect identity and data

    Design access controls and data flows before connecting models to enterprise systems.

  3. 03

    Evaluate the whole system

    Measure retrieval, model output and agent behaviour instead of treating the model as an isolated component.

  4. 04

    Test realistic failure modes

    Test for prompt injection, sensitive data exposure, unsafe tool use and excessive agency.

  5. 05

    Keep people accountable

    Use monitoring and human review so engineering responsibility remains clear in production.

Currently

Career overview

Experience and study, side by side

Working Studying

I have completed four degrees while working full time and am now completing a Master of Research. Formal study has accompanied my career rather than simply preceded it.

Explore the site

Experience

AI and security engineering across enterprise AI, regulated banking and telecommunications platforms operating at high volume.

Studies

Formal depth in AI, cyber security, business and computer engineering, with current research at Macquarie University.

Open source

Selected tools and experiments in model routing, cost analytics, retrieval-augmented generation and MCP.

About

How I combine AI engineering, security, technical leadership, research and mentoring.