IRL · Projects

MOOT — Many Optimization Tasks

A curated benchmark of 120+ multi-objective optimization tasks from real software engineering and systems research. Replaces toy benchmarks with real high-dimensional problems.

MSR'26 A Repository of Many Multi-Objective Optimization Tasks [pdf]
Code: github.com/timm/moot

Explanation — causality, stability, trust

Lead: Amirali Rayegan. Can causal methods make software analytics more stable, interpretable, and trustworthy than correlation-based ones?

EMSE'25 Causal Graphs in SE [pdf]
JSS'26 Explaining Optimization Heuristics [pdf]
ICSE'26 Shaky Causal Structures talk [pdf]

Optimization — simple beats complex

Lead: Kishan Kumar Ganguly. Simple, sample-efficient optimizers routinely beat complex ones on SE problems, because SE data collapses to a few buckets (the "BINGO" effect).

SSBSE'26 Zoom, Don't Wander: Regional Search vs. Pareto vs. Global Optimization
arXiv'25 BINGO! Simple Optimizers Win Big [pdf]
FSE'26 How Low Can You Go? [pdf]
Code: Bingo · DataCentricFuzzJS

Agentic Systems — LLMs as fast typists

Lead: Srinath Srinivasan. How far can we push lightweight, LLM-powered agents on SE tasks without drowning them in compute?

TOSEM'26 LLMs Warming-Up Active Learning [pdf]
EMSE'26 Improving LLM Annotations [pdf]
SmartOracle: agentic approach to differential oracles (preprint)

Testing — coverage isn't enough

Lead: Kishan Kumar Ganguly. Why are some inputs more likely to crash JS engines? Move beyond coverage to data-centric fuzzing that targets the causes of failure.

IST'26 From Coverage to Causes: Data-Centric Fuzzing for JS Engines [pdf]