Technical Program Manager · AI Systems Builder
Scaling IdeasInto AI Systems.
I run programs at Microsoft Azure scale — developer platforms, QaaS infrastructure, AI rollouts across 12,000+ engineers. Then I go home and build working AI systems from scratch.
Latest: DEXter, a real-time voice agent with sub-400ms measured latency, built on WebRTC and the Google Live API. Before that: $60M+ in government SaaS contracts, consolidated testing platforms, and a phased AI rollout to 12K+ Copilot Studio users.
Case Studies.
Enterprise AI products, platform orchestration, and scalable systems — built at Microsoft Azure, Wipro, and GovTech scale.
Enterprise SaaS · Microsoft Azure
QaaS 2.5 Orchestration
Consolidated 5 fragmented testing services into a unified orchestration layer serving 500+ engineers at Microsoft Azure — fastest adoption in team history.
Wipro · Embedded at Microsoft Azure
AI Juno & Vega Labs Portal
Unified developer portal merging LLM-powered AI Juno assistant with Vega Labs lab scheduling — serving 12,000+ Azure engineers globally.
DEXter
MCP_SUITE
v1.0 · Protocol Layer
Unified Model Context Protocol interface bridging DEXter Rise, agentic workflows, and system backends — with constitutional safety constraints at protocol level.
// Initialize MCP v1.0 Handshake
const mcp = new DEXterMCP({ version: '1.0' });
mcp.connect({ backend: 'dexter-rise' });
mcp.applySafety({ alignment: 'Constitutional' });
✓ Protocol handshake complete
// Engineering in progress...
▋Writing Archive.
Build notes, architecture decisions, and technical deep dives.
Bypassing Voice Wrappers for Raw Google APIs
A builder's guide to sub-400ms conversational bots: architecting WebSocket audio pipelines directly on Google, avoiding wrappers, and engineering edge-based steering to keep models on track.
Local Token Dynamics & Context Optimization
An engineering look at tokenization quirks: why byte-pair encoding boundaries break regex filters, how to budget local vs. cloud context windows, and structural chunking tradeoffs in retrieval-augmented pipelines.
The Genesis of Dexter
A Complementary Local-First Architecture for Targeted Data Safety
A structural deconstruction of the air-gapped, hardware-rooted MCP engine built to eradicate corporate telemetry capture, administrative friction, and cloud dependency from high-stakes compliance operations.
Zero Telemetry
Air-gapped runtime
Native Speed
Memory-mapped exec
Autonomous Ops
No cloud dependency
Local LLM
Open-weights on-device
Get in Touch.
I'm open to technical program leadership, AI systems engineering, and developer platform collaborations. Let's build something rare.