SShortSingh.

Programming

0
ProgrammingDEV Community ·

Developer Builds Multi-Tenant RAG Knowledge Base with Source Citations and Demo Mode

A full-stack developer at RA Technologies has built KnowBase AI, a multi-tenant SaaS knowledge base that lets businesses upload documents and receive AI-generated answers grounded in their own content. The system uses a retrieval-augmented generation pipeline covering document ingestion, intelligent chunking with overlap, semantic retrieval, and source-cited responses streamed via server-sent events. Multi-tenancy required isolating each workspace's data at the retrieval layer to ensure one tenant's documents never surface in another's answers. The tech stack includes Next.js 16, TypeScript, SQLite with Prisma, and a provider abstraction layer supporting OpenAI, Gemini, and Claude, with a mock fallback enabling a public demo without API key exposure. The developer highlights source citations as the critical feature that makes the tool trustworthy for real business use cases like customer support.

0
ProgrammingDEV Community ·

Google indexed 13 of 16 pages on a brand-new site within one morning

A new website with no backlinks, no audience, and no prior history was launched with 16 pages, a sitemap, and a connected Google Search Console property. By the following morning, 13 of the 16 pages had been indexed by Google without any paid service or special intervention. The team had used Search Console's Request Indexing feature on 9 of the pages that first afternoon, while the remaining 6 were left untouched, resulting in 9 out of 9 requested pages and 3 out of 6 unrequested pages being indexed. The data suggests organic indexing can happen quickly for new sites, though the small sample size prevents any firm conclusions about whether the Request Indexing button made a meaningful difference. The findings also highlight that Google's Indexing API is officially intended only for job postings and live-stream events, meaning its use for general content may not guarantee indexing action despite returning a success response.

0
ProgrammingDEV Community ·

How a Developer Uncovered a Fake Digital Identity Haunting Open-Source Spaces

Senior backend engineer Sean Byrne began investigating after a colleague mistakenly contacted a GitHub account sharing his name, photo style, and bio. The suspicious profile had no followers, no activity, and repositories set to private, yet references to a 'Sean Byrne' appeared across forums, mailing lists, and open-source documentation. A 2026 Hacker News thread detailing the same investigation went viral, with dozens of users reporting similar phantom identities credited in real projects. OSINT analysis revealed a pattern of high-quality, credibility-building contributions spread thinly across platforms, with no signs of genuine long-term engagement. The fake persona appeared deliberately constructed to lend legitimacy to code that no real person had authored or reviewed.

0
ProgrammingDEV Community ·

Developer Builds AI Disaster Assessment and Women's Safety Tools at Gen AI Academy APAC

A participant in the Gen AI Academy APAC Edition has shared insights from the ongoing program, which combines artificial intelligence with hardware development. As part of HackTronix 2.0, the developer is working on a 3D LiDAR-based system designed to evaluate shelter damage following natural disasters. A second project, called EVA (Early Vulnerability Assessment for Women's Aspirations), aims to build an AI platform supporting underrepresented groups. The author highlights that real-world AI development requires understanding hardware constraints and deployment challenges beyond theoretical knowledge. The post encourages others to participate in similar hackathons and bootcamps, emphasizing hands-on learning over credentials.

0
ProgrammingDEV Community ·

Developer Builds AI Thinking Partner After Realising His Writing Block Was a Clarity Problem

A developer created ThoughtForm, an AI-powered tool, after recognising that his difficulty writing was not a writing problem but a coherence problem — his ideas were formed but disorganised. Inspired by the Socratic method, the tool acts as a conversational thinking partner, asking questions to help users examine, refine, and articulate their ideas before committing them to text. Unlike conventional AI writing assistants, ThoughtForm is designed to produce the user's own words and arguments rather than generate content on their behalf. The developer argues that the real value of such a tool lies in structured thinking and clarity, not in producing polished prose for an audience.

0
ProgrammingDEV Community ·

Developer Explains Portfolio Site Architecture Using a Shopping Centre Analogy

A developer named Adam Belton has written about the architecture behind his personal portfolio website, AdamBelton.com, which hosts his writing and upcoming projects. The site uses a ports-and-adapters architecture comprising host, delivery, and application layers, along with capabilities, ports, adapters, and composition. To make the technical structure more accessible, Belton uses a shopping centre metaphor, where different units represent distinct projects that share common resources such as security, storage, and AI capabilities. One of those projects, ThoughtForm, is a conversational AI workspace designed to help users explore, organise, and articulate their thoughts through guided questioning and idea mapping. ThoughtForm has not yet launched, but Belton plans to release it as his first demo product in the near future.

0
ProgrammingHacker News ·

Cambridge professor Jason Arday, who resigned over plagiarism claims, found dead

Jason Arday, a professor at Cambridge University, has been found dead following his resignation amid plagiarism allegations. The news was reported on August 14, 2026, by The Guardian. Arday had been at the centre of a high-profile academic misconduct row at Cambridge. His death comes shortly after he stepped down from his position at the university in connection with the plagiarism claims.

0
ProgrammingDEV Community ·

Why AI-generated tests can be misleading and spec-first TDD matters

A software engineering perspective published on DEV Community argues that AI-generated tests written after code is produced can create a false sense of reliability. The core concern is that when AI writes both the code and the tests, the tests may simply validate the AI's own assumptions rather than genuine requirements. The author advocates for a spec-first Test-Driven Development approach, where specifications and tests are defined before any code is generated. This method is presented as essential for maintaining meaningful quality assurance in AI-assisted engineering workflows.

0
ProgrammingDEV Community ·

Flock Safety cuts license-plate data retention to 7 days after misuse scandal

Flock Safety CEO Garrett Langley announced on August 13 that the company is reducing its default license-plate-reader data retention period from 30 days to seven, following a Washington Post investigation that found nearly 50 officers accused of running unauthorized searches. The company also plans to make misuse detection mandatory and require a case code on every search by year-end. Flock's own data reveals that over 90% of searches are completed within a week, meaning the additional 23 days of retention served fewer than 10% of queries across its 120,000-camera network. Critics including the ACLU and the Institute for Justice dismissed most of the announced measures as insufficient, arguing they address detection rather than the underlying problem of unchecked police access. Civil liberties advocates contend that search decisions should require judicial oversight rather than remaining at officers' discretion.

0
ProgrammingHacker News ·

GCC Nested Functions Explored Using Wide Pointers Without Trampolines

A technical article published on July 14, 2026, examines an advanced use case of GCC's nested functions feature. The piece focuses on implementing nested functions using wide pointers while avoiding the use of trampolines, which are typically required for such functionality. Trampolines are small executable code stubs that GCC normally generates on the stack to handle nested function calls. The author explores an alternative approach that sidesteps this mechanism, which has implications for security and portability in C programming.

0
ProgrammingDEV Community ·

How Chaotic Math Functions Are Being Used to Encrypt Images Securely

Researchers and developers are exploring chaos-based encryption as a more effective alternative to traditional ciphers like AES for securing digital images. Chaotic systems, such as the logistic map, generate sequences that are deterministic but highly sensitive to initial conditions, making them suitable as pseudo-random number generators for cryptographic use. Images pose unique challenges for conventional encryption due to high pixel redundancy and strong spatial correlations, which chaos-based schemes address through a two-stage process of pixel scrambling and value substitution. Modern implementations often chain multiple chaotic maps or use higher-dimensional systems like the Lorenz attractor to expand the key space and eliminate predictable patterns. This combined permutation-diffusion approach offers both speed advantages and stronger resistance to statistical attacks compared to block cipher methods.

0
ProgrammingDEV Community ·

Developer Finds Solo AI Outperforms Multi-Agent Setup in Coding Orchestrator Tests

A developer built Sol-Luna Orchestrator, an open-source orchestration layer for OpenAI Codex, designed to let a supervisor AI decide whether to delegate tasks to worker agents rather than automatically splitting all work. The system uses GPT-based models in two roles: a supervisor called Sol that owns task planning and review, and worker instances called Luna that are only invoked when delegation is deemed worthwhile. Contrary to expectations, benchmarks consistently showed a single Sol agent working alone was faster than parallel Luna workers, even on tasks deliberately designed with independent workstreams. The project introduced practical safeguards for multi-agent coding, including isolated Git worktrees, file-scope enforcement, and independent verification of worker outputs. The key finding was that the optimal number of workers frequently turned out to be zero, suggesting coordination overhead can outweigh the benefits of parallelism for many coding tasks.

0
ProgrammingDEV Community ·

Stolen Credentials, Not Hacking, Now Drive Most Cloud Storage Breaches

Cloud storage security is increasingly undermined by identity-based attacks rather than infrastructure exploits, with stolen credentials serving as the initial access vector in roughly 22% of confirmed breaches according to Verizon's 2025 Data Breach Investigations Report. Automated tools continuously test stolen login combinations against cloud services, a technique known as credential stuffing, making reused passwords a critical liability. Researchers analyzing a 19-billion-password leak found reused or duplicated credentials were widespread, lowering the bar for attackers significantly. Configuration errors compound the risk, as misconfigured public-facing storage buckets can expose sensitive data without requiring any login at all. Security experts emphasize that under the cloud shared responsibility model, customers — not providers — are accountable for access controls, making multi-factor authentication and proper permissions essential defenses.

0
ProgrammingDEV Community ·

Day 27: Git Revert vs Reset and Making an AWS VPC Subnet Public

A DevOps learning exercise on Day 27 tackled two deceptively simple-sounding tasks: undoing a pushed Git commit and making an AWS VPC subnet publicly accessible. Git's revert command safely undoes a pushed commit by adding a new inverse commit to the history, while reset rewrites history and should never be used on already-pushed commits shared with others. On the AWS side, a 'public subnet' is not a built-in toggle but requires four distinct steps: creating an Internet Gateway, attaching it to the VPC, adding a default route in a route table, and associating that route table with the subnet. Missing any single step produces no error message — just silent connectivity failure, making it particularly difficult to debug. Both tasks illustrate how operations that appear to be a single action are in fact multi-step processes with meaningful consequences if done incorrectly.

0
ProgrammingDEV Community ·

Developer Builds Hinglish Voice AI Counselor for Indian Students in 10 Days

A developer created EduGuideAI, a low-latency conversational voice assistant aimed at helping Indian students and parents navigate educational queries in code-mixed Hindi-English. The project was built in 10 days as part of Murf AI's VoiceForBharat challenge, using LiveKit for real-time audio, Murf Falcon for text-to-speech, and an LLM with academic guardrails for reasoning. The system keeps response latency under 800 milliseconds and can retrieve live course details, remember returning users' context, and hand off complex cases to human counselors. To handle Indian conversational patterns, the team fine-tuned voice activity detection to distinguish brief acknowledgments like 'haan' or 'okay' from actual interruptions. The codebase has been made publicly available on GitHub under the murf-livekit-starter repository.

0
ProgrammingDEV Community ·

Common Data Privacy Myths That Are Now Driving Real Legal Risk

A DEV Community analysis highlights widespread misconceptions about data privacy that are increasingly surfacing in regulatory actions and class-action lawsuits in 2026. The 'nothing to hide' argument is identified as fundamentally flawed, since companies routinely combine legal behavioral data — such as browsing history and location trails — to build predictive profiles used for targeting and persuasion. The Cambridge Analytica scandal is cited as a landmark example of how ordinary, lawful digital activity can be harvested to influence people's decisions. On cookie consent, the piece clarifies that GDPR does not mandate banners specifically, but requires genuine informed consent, which many current banner designs fail to provide. Regulators, including the UK's Information Commissioner's Office, are now actively scrutinizing the top websites for non-compliant consent practices.

0
ProgrammingDEV Community ·

Passwords Alone Leave Cloud Storage Vulnerable to Modern Credential Attacks

Compromised identities now account for over 70% of cloud breaches, while human error contributes to 88% of all data breach incidents, highlighting the limits of password-only security. Attackers increasingly rely on phishing, credential stuffing, and infostealer malware rather than technical exploits, since stolen credentials allow them to log in as legitimate users without triggering alerts. Cloud storage systems are high-value targets because they hold financial records, source code, and customer data central to business operations. Multi-factor authentication, particularly app-based TOTP or hardware security keys, significantly reduces the risk by ensuring a stolen password alone cannot grant account access. SMS-based verification offers some improvement but remains susceptible to SIM-swapping, making stronger MFA methods the more reliable choice for protecting cloud environments.

0
ProgrammingDEV Community ·

Lean Founder on AI, Formal Verification, and the Future of Handwritten Math

In a wide-ranging interview, the creator of Lean — a system that functions as both a programming language and a proof assistant — discussed how formal verification can guarantee the absence of bugs, going beyond what traditional software testing allows. Lean is built on dependent type theory, enabling users to write code, define properties about that code, and produce machine-checkable proofs. The founder noted that mathematical verification tends to involve small statements with deep proofs, while software verification deals with large systems and shallower but structurally complex proofs. He highlighted a future where AI-generated code optimizations come paired with formal proofs of correctness, fundamentally changing how software is delivered. Lean, now backed by a nonprofit foundation and AWS funding since 2023, is being steered toward broader software verification use cases alongside its established role in mathematics.

0
ProgrammingDEV Community ·

SEO Specialist Builds Lightweight Client-Side Word Counter Tool Using AI Assistance

An SEO specialist shared a personal project on DEV Community: a real-time, client-side word counter tool built with AI-assisted development. The tool was created to replace bulky third-party text analysis sites and runs entirely in the browser using pure JavaScript and PHP. It offers real-time counts for words, characters, sentences, and paragraphs, along with estimated reading and speaking times. The developer addressed technical challenges such as regex handling for hyphenated and multi-language words, as well as DOM performance optimization for documents exceeding 10,000 words. The project has been shared publicly, with the developer inviting feedback on bugs, features, and code optimization.

0
ProgrammingDEV Community ·

APIMart Launches Discounted AI API Aggregator Supporting GPT-5 and Sora 2

APIMart is a new AI API aggregator platform that offers discounted access to multiple AI services including GPT-5 and Sora 2. The platform targets developers and web professionals looking to integrate AI APIs at reduced costs. It was highlighted on DEV Community on August 14 by a contributor named Anthony Max. The service appears aimed at simplifying and lowering the barrier to accessing multiple AI APIs through a single aggregated platform.

← NewerPage 80 of 1298Older →