The résumé.

Member of Technical Staff · Engineer #7

Midstream · San Francisco

• Leading document extraction and contract ingestion for 8+ health systems with $100B+ in combined net patient revenue.
• Working on post-training models for document extraction.
• Building an extraction harness with execution traces and human feedback to curate gold-standard datasets.
• Developing evaluation benchmarks and iterative improvement loops across document types.

Founding engineer · Engineer #1

Pipelines · San Francisco

• Built a full-stack AI R&D platform for multimodal data collection, LLM generation, and model evaluation.
• Built visual DAG workflows, state machine orchestration, MCP tool integration, and multi-tenant access control.
• Developed programmatic, LLM-as-judge, and human evaluation with versioned criteria and synthetic data quality gates.
• Architected asynchronous workflow chains and specialized worker pools; deployed on GCP with Cloud Run and Firebase.

Software Engineer

Meta · Menlo Park

• Built AI integrations, machine learning improvements, and product infrastructure across Meta Ray-Ban and Instagram.
• Integrated on-device voice and NLP with the Dexcom API for real-time glucose insights and autonomous reminders.
• Owned Instagram’s incremental revenue reporting pipeline attributing $3B+ in annual revenue, with a self-serve version used by 1,000+ engineers.
• Improved follow prediction accuracy and stability by 21% through training data changes and a weighted Follow Quality Score.
• Developed an agentic system to analyze codebases, trace logic, and generate workflow automations from natural language.
• Built a React ad portal and fixed 16% impression under-reporting; led the Reels Ads Follow Button launch with a 0.42% lift in ad score.

Software Engineer

Bloomberg · New York

• Built a troubleshooting logging system recording 10 million daily queries and responses in AWS S3 through a Kafka consumer.
• Built a data quality validation tool for a backfill worker.

Data Specialist

Sandvik Asia · Bengaluru

• Owned the Austria division’s digital services solution, from data architecture and implementation to reporting with MySQL, Azure Analysis Services, and Power BI.
• Built a decision tree model on live telemetry and maintenance data to predict loader breakdowns, reducing downtime by 15%.

MS, Computer Science

University of Massachusetts Amherst

BE, Computer Science and Engineering

R.N.S. Institute of Technology (VTU)