PDF Tools: Checking the Download, Not Just the Preview Englisch
An engineering case study of independent export validation, browser resource cleanup, and bounded privacy claims in PDF Tools.
Artikel
Technische Artikel von Akshat Gupta über Large Language Models, generative KI, NLP, ML-Sicherheit und angewandte KI-Entwicklung.
Memory in Large Language Models Englisch
A guide to the mechanisms called LLM memory: model parameters, context, KV caches, retrieval, recurrent state, and persistent application storage.
LLM Agents: From Model Output to Reliable Action Englisch
A guide to reliable LLM agents: typed tools, permissions, retries, human approval, observability, security, and trajectory-level evaluation.
Fine-Tuning LLMs with LoRA Englisch
A practical guide to deciding when an LLM needs fine-tuning, understanding LoRA and QLoRA, and building an evaluated PEFT training workflow.
Speaker Anonymization: Privacy Beyond Changing a Voice Englisch
A threat-model-first guide to speaker anonymization: system architectures, informed attackers, privacy and utility metrics, and lessons from SpeakerDiff.
Model Extraction Attacks: Threat Models and Defences Englisch
A threat-model-first guide to behavioural cloning, parameter recovery, query strategies, measurement, detection, watermarking, and layered defences.
Understanding Attention in Transformers Englisch
A concise technical guide to transformer attention: masks, position, efficient kernels, KV caches, and interpretability limits.
Building Reliable RAG Systems Englisch
An end-to-end guide to building and evaluating production RAG systems, from document ingestion and hybrid retrieval to grounded generation and observability.
How Do You Evaluate LLM Systems? Englisch
A production-oriented framework for evaluating LLM applications across model quality, retrieval, agent behaviour, safety, latency, and cost.
Prompt Design for Production LLM Systems Englisch
A production guide to prompt contracts, structured outputs, context construction, tool use, evaluation, observability, and prompt-injection defence.
Knowledge Graphs: Data Models, Construction, and Graph RAG Englisch
A practical guide to knowledge-graph models, identity, provenance, construction, querying, evaluation, and their role in retrieval-augmented generation.
What Are Diffusion Models? Englisch
A technical guide to diffusion models: the forward process, noise-prediction objective, reverse sampling, guidance, latent diffusion, and practical trade-offs.
Ontologies in Knowledge Graphs: Reasoning and Validation Englisch
A practical guide to ontology design in knowledge graphs, including RDF, RDFS, OWL, open-world reasoning, SHACL validation, and production trade-offs.
Vector Databases: Indexes, Retrieval, and Production Trade-offs Englisch
A guide to vector search, ANN indexes, filtering, hybrid retrieval, production operations, and deciding whether a dedicated vector database is necessary.
Fairness in Machine Learning: Metrics, Trade-offs, and Evaluation Englisch
A practical guide to defining, measuring, and monitoring fairness in machine-learning systems without hiding the policy choices behind one metric.