Aarush Kumbhakern

Visiting Research Student — University of Sydney

Advisor: Prof. Rahul Gopinath · remote

• Designed and built the experimental harness and full empirical evaluation of drdd, a structure-agnostic drop-in replacement for ddmin that preserves its 1-minimality guarantee while deferring restarts to the single-element level; measured a 2–60× reduction in oracle calls across four input formats (ffmpeg, XML, binutils, crashjs) at matching reduction quality. First-author paper under major revision at ISSRE 2026. code ↗
• Built grammar-based fuzzers in C and compiler-style tooling that generates them from grammar specifications, iterating on profiling and low-level implementation design for performance. code ↗
• Current direction: inferring acceptance grammars from ddmin runs to generate minimal inputs.

Undergraduate Thesis — HyDRA, Ashoka University

Advisor: Prof. Bhargab B. Bhattacharya

• Recast a sample-preparation capstone as an architecture-and-abstraction problem: designed HyDRA, a programmable slug-flow microfluidic substrate with an SSA, integer-valued ISA over discrete fluid operations, plus its micro- and physical architecture.
• Specified a finite-state chip abstraction whose safety properties reduce to Boolean combinations of integer linear inequalities — decidable ahead of time in polynomial resources.
• Framed the result as a programmable, statically verifiable substrate (architecture + ISA + static guarantees); accepted at ISVLSI 2026. thesis ↗

Undergraduate Researcher, Food Computing — Mphasis AI & Applied Tech Lab

Advisor: Prof. Partha Pratim Das

• Designed a domain-specific language that models cooking procedures as action graphs — ingredients, intermediate components, transformations, dependencies, and procedural branches — turning loose natural-language recipes into structured computational objects; published at MMFood ’25.

Engineering Intern — Catalis Packaging Technologies

Vadodara, India

• Built a suite of interconnected Linux/systemd services and daemons on Raspberry Pi-class hardware — a USB-triggered automated workflow for industrial HMIs, with machine telemetry collected over Modbus and streamed to the cloud — owning the end-to-end architecture and operational reliability. code ↗

Three papers

HyDRA (ISVLSI 2026, oral); Dr. DD (under major revision, ISSRE 2026); cooking-procedure ontology (MMFood ’25, ACM). Full list →

Best Undergraduate Thesis

Department of Computer Science, Ashoka University

Engineering International Scholar Award — University of Toronto

Entrance scholarship · CAD $140,000 · declined

Teaching Assistant — Lodha Genius Program, Ashoka University

• Mentored five students one-on-one in data interpretation and simulation development for an intensive course on computational modelling and biological-systems analysis.
• Delivered a lecture on version control and reproducible-research workflows.

C, Python, JavaScript

Linux, systemd, Git, Make, Modbus, profiling & benchmarking, experimental harnesses

Compilers & DSLs, program analysis, fuzzing, delta debugging & input reduction, empirical evaluation, finite-state modelling & static reasoning, scheduling & optimization, embedded & cyber-physical systems, machine learning

Theory of Computation · Programming Languages and Translation · Computer Organization and Systems · Design and Analysis of Algorithms · Numerical Algorithms and Optimization · Information Security · Computer Networks · Probability and Statistics · Reinforcement Learning · Machine Learning

Physiology · Cell Biology · Chemical Basis of Life · Force and Motion in Biology (biomechanics)

Available on request.