Source author record

Parth Thakkar

Parth Thakkar appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Artificial Intelligence Databases Distributed, Parallel, and Cluster Computing Performance Software Engineering

Catalog footprint

What is connected

2works

5topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2026arXiv

ProgramBench: Can Language Models Rebuild Programs From Scratch?

Turning ideas into full software projects from scratch has become a popular use case for language models. Agents are being deployed to seed, maintain, and grow codebases over extended periods with minimal human oversight. Such settings require models to make high-level software architecture decisions. However, existing benchmarks measure focused, limited tasks such as fixing a single bug or developing a single, specified feature. We therefore introduce ProgramBench to measure the ability of software engineering agents to develop software holisitically. In ProgramBench, given only a program and its documentation, agents must architect and implement a codebase that matches the reference executable's behavior. End-to-end behavioral tests are generated via agent-driven fuzzing, enabling evaluation without prescribing implementation structure. Our 200 tasks range from compact CLI tools to widely used software such as FFmpeg, SQLite, and the PHP interpreter. We evaluate 9 LMs and find that none fully resolve any task, with the best model passing 95\% of tests on only 3\% of tasks. Models favor monolithic, single-file implementations that diverge sharply from human-written code.

preprint2021arXiv

Scaling Hyperledger Fabric Using Pipelined Execution and Sparse Peers

Permissioned blockchains are becoming popular as data management systems in the enterprise setting. Compared to traditional distributed databases, blockchain platforms provide increased security guarantees but significantly lower performance. Further, these platforms are quite expensive to run for the low throughput they provide. The following are two ways to improve performance and reduce cost: (1) make the system utilize allocated resources efficiently; (2) allow rapid and dynamic scaling of allocated resources based on load. We explore both of these in this work. We first investigate the reasons for the poor performance and scalability of the dominant permissioned blockchain flavor called Execute-Order-Validate (EOV). We do this by studying the scaling characteristics of Hyperledger Fabric, a popular EOV platform, using vertical scaling and horizontal scaling. We find that the transaction throughput scales very poorly with these techniques. At least in the permissioned setting, the real bottleneck is transaction processing, not the consensus protocol. With vertical scaling, the allocated vCPUs go under-utilized. In contrast, with horizontal scaling, the allocated resources get wasted due to redundant work across nodes within an organization. To mitigate the above concerns, we first improve resource efficiency by (a) improving CPU utilization with a pipelined execution of validation & commit phases; (b) avoiding redundant work across nodes by introducing a new type of peer node called sparse peer that selectively commits transactions. We further propose a technique that enables the rapid scaling of resources. Our implementation - SmartFabric, built on top of Hyperledger Fabric demonstrates 3x higher throughput, 12-26x faster scale-up time, and provides Fabric's throughput at 50% to 87% lower cost.