• Drive AI inference efficiency for compute-heavy ML serving focused on systems and hardware levers including compiler optimization, CPU efficiency, and hardware enablement.

• Led compiler tuning experiments exploring LLVM march/mtune configurations across production inference services; characterized the performance ceiling and mapped the structural library dependencies that gate next-generation hardware enablement.

• Currently building automation for new hardware validation and expanding the range of ML workloads that can run on shared CPU compute capacity.

• Partner across compiler, runtime, and hardware teams on multi-million-dollar inference cost reduction goals.

• Owner of distributed data lifecycle systems processing 12+ exabytes across all Microsoft

365 enterprise and consumer workloads

• Designed and scaled distributed deletion pipelines purging 1.3 EB/month of

user‐deleted data across worldwide farms, using Azure Job Queues in a

producer‐consumer pattern with redundant safety systems ensuring zero

accidental data loss

• Architected high‐throughput compression workflows delivering

$4.5M/month in cost savings across 125 PB/month, including backlog

processing of 211 PB of previously uncompressed data

• Built distributed orphan collection system using fault‐tolerant protocols

across geo‐distributed clusters, recovering $2.9M in storage costs (80 PB

total)

• Designed account lifecycle deletion pipeline achieving $1.8M/month

savings (50 PB/month) with zero data loss across billions of objects

• Redesigned data integrity scanning using distributed priority queue

architecture, improving SLA coverage from 86.81% → 99.86%

• Engineered fault‐tolerant failover systems for region‐down scenarios,

ensuring high availability for critical data pipelines across multiple

datacenters

• Architected safety‐critical data recovery mechanism achieving 100%

success rate for disaster recovery, safeguarding exabytes of production data

• Led infrastructure efficiency strategy, designing cost models for single‐LRS

deletion and root blob compression that captured $8M+ in annual savings

• Extended data integrity systems to support next‐gen storage platform

migration, ensuring correctness guarantees during live data movement at

exabyte scale

• Scaled org capacity by hiring and mentoring engineers across 4 systems as

service grew 5x

• As the designated AI Coding Champion of the org, leading adoption of

AI‐assisted development and conducting org‐wide seminars

• Developed full-stack ReactJS based web applications which processed, analyzed, and rendered user data visually

• Managed time-sensitive updates, including content changes and database upgrades to MongoDB

• Implemented user impersonation feature for a template Angular application

• Migrated from using sessions to JWT for the authentication

• Incremental development of a Python-based, API-driven test automation tool for Redfish®

• Developed a caching mechanism to store HTTP GET requests using JSON serialization, resulting in a ≈ 50x speed improvement over the existing tool

• Spearheaded the transition from SharePoint platform to MERN stack, reducing latency by 40% and increasing database admin effectiveness

• Rebuilt the tool’s log representation feature using Angular and Material Design, which led to the easier identification of assertion failures