Human-AI collaboration
Collective action in shared supply chains
Co-developing a working paper and online research project on group decision-making in shared supply chains.
Computer Science at NYU Abu Dhabi
I build AI systems and test what they change.
My work combines model adaptation, human-AI experimentation, and practical machine-learning systems.
I turn questions into datasets, training pipelines, and experimental systems, then evaluate whether a result is stable, interpretable, and useful beyond a single run.
B.S. expected May 2027
Research engineering and evaluation
Selected work
Questions with artifacts
A few projects that show how I work: define a question, build the system or study, and make the evidence inspectable.
Model adaptation
Bio-Informed LoRA for signal peptide prediction
Implemented and evaluated an ESM-2 adaptation method with a three-seed benchmark and exploratory follow-up checks.
Reproducible evaluation
Hinglish named-entity recognition benchmark
Fine-tuned multilingual models and documented zero-shot baselines for code-mixed named-entity recognition.
Method
How I approach a problem
01
Frame
Specify the decision, uncertainty, or behavior worth studying.
02
Build
Prepare data and create the model, tool, or experimental system.
03
Test
Use controls, common protocols, and repeated runs to evaluate it.
04
Communicate
Share evidence, limits, and usable artifacts without overstating a result.
Updates
Recent work
- Jun 2026 Prepared a June 2026 working paper with Benjamin Rosche and Hanan Salam on collaboration in shared supply chains. Study implementation is in progress. Project overview →
- 2026 Continued research on parameter-efficient adaptation and careful evaluation of ESM-2 protein models. Research overview →
Focus areas
Methods and applications
Data and statistical analysis
Turning complex and unstructured data into measurements, representations, and evidence that can answer substantive questions.
AI experimentation and evaluation
Designing controlled comparisons and using statistical inference to evaluate model behavior, efficiency, and downstream outcomes.
Applied systems and public impact
Building reproducible computational tools and connecting technical results to decisions, human needs, and socially consequential problems.