Showing 12 of 12 projects
Automated Lead Discovery and Website Assessment System for SMEs in the Netherlands
ALDWAS is an automation project that builds and qualifies a nationwide list of business sales prospects for a software services company. The system works in two phases. In Phase 1, it starts from the open Dutch address register (BAG), keeps only non-residential addresses, and then discovers the business at each location, from public sources such as Google Maps and OpenStreetMap, capturing the business name, contact details, categories, reviews and website link. The result is a structured, deduplicated database of businesses across the country. In Phase 2, the system visits each business's website and automatically assesses its quality and age, producing a "modernization score" from technical signals (mobile-friendliness, security, outdated technology, last-updated dates, and historical archive activity). It also extracts a valid business email address where one is available. The output is a ranked, segmented list: businesses with no website at all, and businesses whose sites are clearly outdated. Both segments represent companies with a genuine need for new or improved software, ready for a targeted email outreach campaign. The project is designed to be built and run region by region, so it can start small, prove itself in one area, and scale nationwide.
Automated Test Case Generation for Database Systems via LLM-Driven Requirement Parsing
In industrial DBMS development, unit tests are commonly derived from requirement documents to validate whether newly implemented features conform to their specifications. These documents describe expected functionality, supported SQL behaviours, boundary conditions, configuration options, and user-visible semantics. In current practice, engineers manually inspect the requirements to identify test-relevant behaviours, design the necessary schemas and SQL statements, and encode the expected results to form unit tests.
Low-Latency Authorization Scope Analysis for Cloud Access Control Policies in Synchronous Interfaces
RAM/IAM policies underpin least-privilege enforcement in the cloud. Audit tools like Alibaba Cloud Access Advisor deliver policy authorization scope analysis via synchronous APIs with strict latency SLAs. State-of-the-art SMT-based methods (e.g., Stratified Abstraction) are accurate but suffer prohibitive latency on production policies with wildcards, high-cardinality Action/Resource sets and interleaved Allow-Deny rules. Coarse-grained fast methods generate excessive false positives, while zero false negatives is a non-negotiable safety baseline.
Fast and Reliable Detection of Silent Data Corruption in AI Accelerator Systems
Large-scale AI training and inference increasingly rely on specialized AI accelerators. A critical reliability threat in such systems is silent data corruption (SDC): an operator or training step completes without an exception, yet returns numerically incorrect tensor values. Unlike crashes or timeout failures, SDC may remain hidden until it appears as unstable loss curves, abnormal gradients, degraded model quality, or unexplained divergence. Detecting SDC is therefore a practical software engineering problem for accelerator qualification, deployment screening, regression testing, and incident diagnosis
Agent Memory in the Wild: Reliability, Efficiency, and Evaluation
Large language model (LLM) agents are moving from single-shot conversations to long-horizon work. Support agents follow a customer for months, assistants accumulate preferences, and coding agents revisit one repository across hundreds of sessions. Context windows cannot hold such histories, and wholesale replay is slow, costly, and degrades reasoning, so a dedicated memory layer has become standard agent infrastructure. Memory systems extract salient facts from interactions, consolidate them into vector indexes, temporal knowledge graphs, or paged context, and retrieve them into future contexts.
Solarman
It's a customised home solar service providing platform, people can order abd get info for their home. Customised order, package, calculator and project info and also client acquisition required. Details will be followed when project started.
ABC Zaber
It's a personal portfolio website for a scholar and social activists with dynamic portfolio in Entrepreneurship and business. So article, blog, info and others international standards will be followed for website.
Qadeem
It's a complete e-commerce website and product category will be like craft, home decor, essential decorative, ornaments etc. Except grocery. Related to lifestyle and fashion like aarong...
CRAFT BD
It's an organisation website where Craft product will be sold and training will be provided via workshop and Research articles will.be published.. So Standard features will be accumulated in this customised segments.
APKS Global
It's an international NGO website, registered in.Bangladesh but for impact driven activities and projects want to seek fund and attract donr organisation so need an NGO portfolio website containing It's standard features and categories with sober look.
Bamboo Chick
It's a Company Portfolio Website of Interior design with natural and bamboo curtain manufacturer. So company information, product catalogue, and project Gallery, blog etc. Will be highlight and for order or book appointment there will be two separate query form and something more.
Implement and find a Benchmark Framework for Security Vulnerability Detection in LLM-Generated Code
The code generation capabilities of large language models(LLMs) have emerged as a critical dimension in evaluating their overall performance. However, prior research has largely overlooked the security risks inherent in the generated code. In this work, we introduce SafeGenBench, a benchmark specifically designed to assess the security of LLM-generated. code.