VP of AI & Engineering · Health Tech · Sydney

Kanishka
Mohaia

I build engineering organisations where AI does real work and people stay accountable — from real-time ML at billions of transactions to custom LLMs in mental health and wellbeing.

Get in touchView LinkedIn
Kanishka Mohaia.
  • 5× throughputEngineering throughput at Sonder from AI adoption, a restructure and stronger hiring
  • LLMs in productionCustom-trained and DPO-aligned for a care Copilot
  • Series B → pre‑IPOScaled engineering and ML at Rokt
  • Named inventorMachine-learning patents, U.S. Patent No. 11,776,013

Experience

From global scale to high trust.

Sonder

VP of AI & Engineering · 2025–Present

Health tech · Mental health & wellbeing

  • I lead Sonder’s entire engineering organisation — product engineering, platform, security, mobile, back-end, data, machine learning and the AI platform — and its transformation into an AI-first health tech platform.
  • I launched Sonder’s first production AI Copilot for care specialists, combining custom fine-tuned LLMs (including DPO alignment), agentic RAG, summarisation and automated evaluation, with human oversight built in.
  • I defined our AI-native adoption model. Together with a team restructure and stronger hiring, engineering throughput rose 5×.
  • I revamped the company’s security posture and established responsible AI governance, so our systems are secure, auditable, clinically aligned and trusted at scale.

Rokt

VP of Engineering, AI/ML · 2020–2025

Head of Engineering, Data & Optimisation · 2017–2020

Ecommerce · Ad tech · Marketplaces

  • Partnering with the CTO, I helped scale engineering from 30 to 120+ engineers across Sydney and New York as Rokt grew from Series B to pre-IPO, including its first New York ML team.
  • I built Rokt’s first real-time ML prediction and auction systems, improving marketplace efficiency by 20%, and scaled ML from millions to billions of transactions.
  • I led machine learning, real-time identity and ad selection for the ecommerce ad marketplace at the transaction moment, including first- and second-price auction design.
  • I built the streaming data network behind attribution and ML, standardised ML pipelines on TensorFlow Extended and launched Rokt’s first incrementality measurement platform.
  • I’m a named inventor on ML patents for contextual prediction and content ranking, including U.S. Patent No. 11,776,013.

Earlier

Datalicious (2014–2017), Lead Data Engineer to Head of Engineering: ad tech, attribution and data platforms. Appen (2008–2014), Software Developer: NLP, translation and speech data for clients including Microsoft, Google and Amazon.

Education

University of Sydney, Master of Data Science (Machine Learning). UNSW, Bachelor in Mechatronics. UTS, Graduate Certificate in Finance.

How I lead AI adoption

From AI-enabled to AI-native.

A three-phase model I defined at Sonder and deployed first across the engineering team.

  1. 01

    AI-enabled

    People use LLM assistants interactively to get their work done.

  2. 02

    AI-integrated

    Agents are built into workflows, but work still runs through interactive sessions.

  3. 03

    AI-native

    Autonomous agents do the work and collaborate; people review it. Take the AI away and the work stops.

5×

Increase in engineering throughput, measured by merged commits to production — the combined effect of AI adoption across product and technology, a team restructure and hiring high-calibre engineers.

Capabilities

What I bring to a leadership team.

  • AI-native transformation

    I move teams from AI-enabled to AI-native, where agents do the work and people review it — with throughput you can measure.

  • Scaling engineering organisations

    I guide companies through growth by scaling the systems, the teams and a high-performing culture together.

  • Machine learning & custom LLMs

    I build ML from scratch — architecture, training, fine-tuning and DPO — and take it to production at scale.

Domains

  • Health tech
  • Mental health & wellbeing
  • AI-native engineering
  • LLM fine-tuning & DPO
  • Real-time ML
  • Marketplaces & auctions
  • Ad tech

Speaking

On stage: the trust gap.

Kanishka Mohaia delivering the DataEngBytes 2026 keynote on stage in Sydney.

Keynote · DataEngBytes 2026 · Sydney & Melbourne

The Trust Gap: Why the Next Wave of AI Won’t Be Won on Capability Alone

Why trust, governance and accountability — not model capability — decide which AI makes it into consequential production.

  • Building a Human-in-the-Loop AI Copilot on AWSAWS Summit Sydney 2026
  • ML/AI Panel: Bringing AI to ProductionDataEngBytes 2025
  • Fine-tuning or RAG?DataEngBytes 2025
  • Career lessons for graduating engineersUNSW Engineering
  • Unleashing the Power of Real-Time Machine LearningDataEngBytes 2023
  • Accelerate Machine LearningAWS for Software Companies Podcast · Listen →

Short bio

Kanishka Mohaia is VP of AI & Engineering at Sonder, a health tech company in mental health and wellbeing, leading the entire engineering organisation and the custom LLMs behind its human-in-the-loop AI Copilot. Previously, Kanishka spent nearly eight years at Rokt, most recently as VP of Engineering, AI/ML, scaling engineering and machine learning from Series B to pre-IPO across Sydney and New York. Kanishka holds a Master of Data Science from the University of Sydney, is a named inventor on machine-learning patents and speaks on trust and production AI.

Contact

Let’s talk.

For executive search, advisory, board and speaking enquiries.

kanishka@mohaia.aiLinkedIn ↗