I build and lead the AI platforms behind products used by hundreds of millions — turning frontier GenAI into dependable, governed experiences at Fortune 500 and national scale.
Two decades building for the world's most demanding organisations
Frontier capability is easy to demo and hard to trust. My job is to make AI dependable — for 400 million people, on the cheapest phone, in their own language.
I've spent 21 years at the intersection of data, cloud, and product — from Fortune 500 warehousing and Big-4 banking regulation to today's agentic, multilingual, voice-first consumer AI. That range is deliberate: shipping AI at scale is as much about governance, latency, and organisational alignment as it is about models.
Guardrails, sandboxing, and human-in-the-loop governance designed in from layer one — not bolted on.
A 2-second perceptible signal on 2G matters more than a perfect answer nobody waits for.
Leading 100+ across three continents taught me that the hardest system to align is the org chart.
Flagship programs where I owned the architecture, the roadmap, and the outcome.
Deliver a trustworthy agentic AI assistant to 400M+ subscribers across 22 Indian languages — voice and text — on low-end 2G/3G devices, across an open ecosystem of first- and third-party partners.
Architected a React Native + JDS client with MCP tool orchestration, LiveKit WebRTC encrypted voice, an AG-UI real-time widget renderer, and a 4-layer guardrail framework (partner governance, schema validation, CSP sandboxing, signed OTA bundles) for zero-trust across partners.
Progressive Response Design guarantees a 2-second signal on 2G; JioAirSync OTA ships P0 patches in under 4 hours with auto-rollback. Leading a 17-member CoreUX team through a 3-phase rollout to GA.
A Fortune 500 manufacturer was locked into legacy Informatica MDM — expensive to license and painfully slow, with source onboarding taking 6+ weeks.
Owned vision and delivery of a GenAI-powered MDM platform on AWS: a zero-loss migration of 50M+ golden records across a raw → curated → golden S3 architecture on Spark/EMR, with LLM profiling + RAG + MCP source onboarding and OpenSearch vector entity resolution.
Decommissioned Informatica, redirecting 8+ source systems to metadata-driven pipelines with guardrails (PII masking, prompt-injection prevention, RBAC) — across 6 SAFe PI cycles and a 30+ member team.
ANZ's Global Asset Finance ran on a legacy Teradata warehouse — analyst queries took 45 seconds, APRA regulatory compliance was manual, and delivery spanned a 100+ member multi-vendor ecosystem across three continents.
Led the Teradata-to-BigQuery migration with a 3-zone architecture — partitioning and clustering, 4-tier IAM, Policy Tags for PII, and Authorized + Materialized Views — plus a collections analytics platform for predictive EMI recovery.
Progressed from Senior Business Analyst to Technical Manager over 9 years while orchestrating IBM, Wipro, TCS, Infosys, and Cognizant across 80+ concurrent initiatives.
Malaya is the rare leader who can hold the boardroom and the architecture diagram in the same conversation. On AEVAH, he took an ambiguous, high-stakes GenAI mandate for a Fortune 500 client and turned it into disciplined, on-time delivery — retiring a legacy platform, migrating 50 million records without a single loss, and saving over $1.5M a year. He earns the trust of engineers and executives alike. I would put him in front of any client, anywhere.
I have partnered with many program leaders; few combine Malaya's technical depth with his calm command of delivery. He re-architected our master-data platform around GenAI with a rigor that made the complex feel inevitable — never losing sight of governance, auditability, or the people doing the work. He is, quite simply, one of the most dependable technology leaders I have worked with.
Open to senior AI leadership conversations — GenAI platforms, agentic products, and enterprise data transformation at scale.