Software Engineering Daily

Radu Gheorghe, a software engineer at Vespa who moved from Elasticsearch and Solr consulting into tensor-based retrieval, explains why single-vector similarity is not enough. He talks about chunking and lossy embeddings. He outlines multi-stage retrieval, re-ranking trade-offs, and how tensors with named dimensions enable richer, scalable search for multimodal and real-time systems.

A deep dive into Anthropic’s controversial Mythos security model and its selective rollout. Tales of recent cloud and supply-chain compromises and lessons in secure-by-default design. A look at massive AI infrastructure spending and how cloud, chips, and model labs are becoming intertwined. Hacker News highlights include AI reviving old projects, compact model-training tricks, and the math behind cheating at Tetris.

Fitz Nowlan, VP of AI and Architecture at SmartBear and co-founder of Reflect, brings expertise in AI-native QA and web test automation. He discusses why web UI testing is uniquely hard. He explains BearQ’s multi-agent approach to exploring apps, coordinating tester agents, and managing test data at scale. He also covers how agents infer app structure and where humans fit into the outer loop.

Yuval Shany, a law professor and former UN Human Rights Committee member specializing in international humanitarian law, discusses AI in warfare. He talks about current autonomous features in drones and targeting, accountability gaps in AI-mediated conflict, meaningful human control, testing challenges for non-deterministic systems, and lessons engineers should heed when building consequential AI.

Benny Chen, co-founder of Fireworks AI and former Meta ML infrastructure engineer, shares his work building platforms to serve and fine-tune open-weight models at scale. He discusses custom kernels, speculative decoding for faster code completion, multi-hardware support, reinforcement fine-tuning, and using production traces to evaluate real-world performance.

Manish Kapur, VP of Product Marketing & Developer Relations at Sonar with hands-on enterprise software experience, and Chris Grams, CVP of Corporate Marketing at Sonar and survey/data specialist, discuss Sonar’s State of Code Developer Survey. They cover rapid AI adoption, widespread but cautious use of AI-generated code, verification and static analysis for trust, tool proliferation and shadow AI, and how teams should prioritize verification and skills.

Simba Khadder, AI strategy lead at Redis and FeatureForm co-founder, explains why context is now the core challenge for agentic AI. He breaks down context engines: on-demand retrieval, fresh data, fast access, and evolving memory. They cover materialized views, semantic layers over Redis, async memory compaction, and how teams must change engineering practices for AI-driven systems.

Eric Broda, veteran software practitioner and co-author of Agentic Mesh, focuses on enterprise-ready architectures for multi-agent systems. He discusses turning agents into distributed infrastructure. Topics include orchestration and state management, trust and certification frameworks, observability and explainability, event-driven agent communication, and practical enterprise use cases.

Nic Benders, Chief Technology Strategist at New Relic with 16 years there, explains the shift from instrumentation to AI-driven observability. He discusses combining statistical anomaly detection with LLMs to surface meaningful signals, tackling alert fatigue via automated triage and self-healing, and the new challenges of observing AI systems with metrics like token usage, answer quality, and cost.

Ryan Lloyd, Chief Product Officer at GuardSquare and former developer-tools specialist, discusses mobile app security. He highlights why mobile differs from web, the evolution of reverse-engineering tools, layered obfuscation and runtime protections, API attestation, threat monitoring, and how LLMs are changing attacker capabilities.