Engineering
500K+
lines of code delivered across 700+ commits
Query-engine extensions, evaluation platforms, and benchmarking pipelines · 2026
Data management / AI / LLMs
Connecting data
to intelligence.
I’m a Senior Research Scientist at IBM. I build systems and lead research programs for data management and AI, with a current focus on LLM-powered query processing and evaluation. I also teach databases at Columbia University.
Research & teachingIBM ResearchSenior Research ScientistColumbia UniversityAdjunct Associate Professor
By the numbers
Research
170+
publications
Including 70+ peer-reviewed papers and 38 patent applications
Teaching
4.45/ 5.00
student survey score
Columbia University · APAN5400 Fundamentals of Data Engineering · Fall 2025
Teamwork
275+
collaborators and co-authors
01 / Selected work
Systems and research programs.
Query processing, evaluation,
and event forecasting.
01 / Enterprise Data Lakes
Semantic Query Processing
Developing a query engine that uses LLMs to process enterprise data, with research on result reliability and reuse.
Explore the work
02 / Evaluation for Data & AI
Benchmarking Done Right
Evaluation methods and tools for Text-to-SQL, semantic table interpretation, data discovery, and LLM-powered queries.
Explore the work
03 / Causal Models & Enterprise Risk
Event Forecasting & Scenario Planning
Led a multi-year research program connecting causal knowledge extraction, event forecasting, and neuro-symbolic scenario planning for enterprise risk.
Explore the workEngineering / Technology stack
Engineering across the data and AI stack.
I build query-engine extensions, evaluation frameworks, and applications for running and analyzing experiments.
Data & query engines
- Presto
- Velox
- Apache Spark
- DuckDB
- Apache Flink
- Apache Gluten
Languages & integration
- Python
- C++
- Java
- SQL
- JavaScript
- TypeScript
- JNI
AI & evaluation
- Large Language Models
- embeddings
- agentic Text-to-SQL
- semantic-query evaluation
Applications & delivery
- React
- FastAPI
- Redis
- OpenAPI
- Docker
- CI/CD
AI coding tools
- Claude
- Codex
- Cursor
- IBM Bob
- GitHub Copilot
Query-engine integration
AI query functionality across Presto, DuckDB, Spark with Velox, and Flink, including C++ and Java integration.
Related work ↗Evaluation infrastructure
React and FastAPI applications with Redis-backed experiment queues, LLM integration, and results analysis.
Related work ↗Benchmarking tools
Agentic Text-to-SQL evaluation pipelines, benchmark datasets, and semantic-query tests.
Related work ↗Research perspective
Research in
data management & AI.
- Semantic query processing with large language models
- Evaluation of Text-to-SQL and other AI-based data systems
- Data integration, discovery, and curation
- Semantic methods for linking and interpreting structured data
02 / Selected publications
Selected recent publications.
2026Publication
2026Publication
2026Publication
I design and build data systems, lead research teams, and develop methods for evaluating AI. My work spans enterprise query processing, benchmarking, data discovery, and event forecasting.
2011 – Present
Senior Research Scientist
IBM T.J. Watson Research Center
2026 – Present
Adjunct Associate Professor of Computer Science
Columbia University