Resume

Agentic data fabric platform · San Francisco Bay Area

• Own the platform's architecture end to end: API contracts, identity model, agent execution, data architecture, and Kubernetes delivery for a platform that lets enterprises run AI agents safely against their operational data.
• Defined the contract-first architecture that turns the platform's open-source data and AI subsystems into one governed product: a single identity flowing through every layer, so subsystems stay swappable and every agent action is attributable to a user.
• Led agent execution from design through production: workspace isolation, session recoverability, and policy-governed tool access (MCP) that holds up to enterprise security review.
• Set technical direction for document intelligence: an ingestion and retrieval architecture (multi-modal RAG) that turns heterogeneous document corpora into queryable knowledge, with every answer traceable to its source page, which customers required before trusting AI-generated answers.
• Made the platform installable and operable as a product. The full 15-service stack deploys into a customer's Kubernetes environment in one step, with tenant isolation, single sign-on across every component, and a release-gating integration suite.
• Hire and ramp the platform's engineers, keep every architecture decision written down with the options rejected and why, and run the build-vs-buy evaluation behind each major subsystem.

AI agents and automation consultancy (SEA and GCC) · Remote

• Designed and shipped production agent systems for enterprises in F&B, insurance, distribution, and industrial IoT. The recurring work: grounding answers in verifiable sources, evaluation that catches regressions before customers do, and integration with the systems of record enterprises run on.
• Built a research agent for a major F&B group that synthesizes trend and product research from 100+ sources into briefs its innovation team acts on directly, with source weighting tuned by the team's own feedback.
• Built a benefits-and-claims assistant for a health-insurance provider: identity-scoped retrieval over plan documents and claims systems, so members resolve questions conversationally instead of through the call center.
• Built a natural-language interface to distributed IoT fleets that lets non-technical operators query and control devices safely: permissioned command paths, blast-radius limits, and confirmation on destructive operations.

Health-equity fintech · Chicago, IL (Hybrid)

• Owned member-experience engineering: the AndMore mobile app, web portal, engagement, customer segmentation, recommendations, and LLM integration.
• Took the company mobile, launching iOS and Android apps in 3 months with a team hired and ramped in parallel, opening a new revenue channel.
• Shipped an in-app AI assistant answering members' benefits questions in real time, barcode scanning built on a custom vision model, and a geolocation recommendation engine across hundreds of partner stores.

Enterprise telehealth serving 1M+ patient interactions · Remote

• Led the 8-person product team shipping the enterprise suite (SSO, self-service onboarding, billing, permissions) that unblocked large health-system deals.
• Directed, in parallel, the 4-person infrastructure team's Azure-to-AWS Fargate migration, cutting infrastructure costs by $100K+ annually.

AI, cloud, SaaS, and IoT applications studio · USA, Netherlands, Pakistan, Vietnam

• Led delivery of AI/ML client solutions across four countries while scaling a distributed team of 20+ engineers. The work spanned computer-vision dental shade matching (productized as Matisse.ai), IoT fleet management for a Philippine delivery network built on in-truck telemetry, and real-time transaction anomaly detection for a US convenience-store chain.

Redwood City, CA

• Guided 6 teams (33 engineers) to launch Spark-as-a-Service on Oracle Kubernetes Engine in 12 months; ran the Fortune 500 proofs of concept that shaped the platform for real-time analytics and ML workloads.

Acquired by Oracle · San Jose, CA

• Architected one of the first data lake platforms built on compiled query execution: interactive OLAP directly over open data formats on Apache Spark and Kubernetes, sustaining 10,000+ queries per second.

Autonomous food robotics, now Creator · San Francisco, CA

• Led the 15-engineer team building the real-time control system, firmware, and computer vision for robotic burger assembly, with 20+ units in the field under centralized fleet oversight.

HANA in-memory database core · USA, Germany, Korea, France

• Engineered distributed query processing and in-memory vectorized execution in C++ inside the HANA core, including a lockless persistent hash table that improved parallel query throughput 200%, with hot paths tuned using SIMD and RDMA.
• Led HANA's Hadoop integration across engine teams in four countries, bringing external data into the distributed query planner.

Thin-film solar manufacturing · Fremont, CA

• Built real-time process-monitoring software for thin-film solar cell production lines, instrumenting manufacturing steps and surfacing live yield and cycle-time metrics that drove the fab's process improvement initiatives.