Goto

Collaborating Authors

 principal software engineer


In AI Sweet Harmony: Sociopragmatic Guardrail Bypasses and Evaluation-Awareness in OpenAI gpt-oss-20b

Durner, Nils

arXiv.org Artificial Intelligence

We probe OpenAI's open-weights 20-billion-parameter model gpt-oss-20b to study how sociopragmatic framing, language choice, and instruction hierarchy affect refusal behavior. Across 80 seeded iterations per scenario, we test several harm domains including ZIP-bomb construction (cyber threat), synthetic card-number generation, minor-unsafe driving advice, drug-precursor indicators, and RAG context exfiltration. Composite prompts that combine an educator persona, a safety-pretext ("what to avoid"), and step-cue phrasing flip assistance rates from 0% to 97.5% on a ZIP-bomb task. On our grid, formal registers in German and French are often leakier than matched English prompts. A "Linux terminal" role-play overrides a developer rule not to reveal context in a majority of runs with a naive developer prompt, and we introduce an AI-assisted hardening method that reduces leakage to 0% in several user-prompt variants. We further test evaluation awareness with a paired-track design and measure frame-conditioned differences between matched "helpfulness" and "harmfulness" evaluation prompts; we observe inconsistent assistance in 13% of pairs. Finally, we find that the OpenAI Moderation API under-captures materially helpful outputs relative to a semantic grader, and that refusal rates differ by 5 to 10 percentage points across inference stacks, raising reproducibility concerns. We release prompts, seeds, outputs, and code for reproducible auditing at https://github.com/ndurner/gpt-oss-rt-run .


Principal Software Engineer (BI Developer) at Eurofins - Bengaluru, India

#artificialintelligence

Eurofins Scientific is an international life sciences company, providing a unique range of analytical testing services to clients across multiple industries, to make life and the environment safer, healthier and more sustainable. From the food you eat to the medicines you rely on, Eurofins works with the biggest companies in the world to ensure the products they supply are safe, their ingredients are authentic and labelling is accurate. Eurofins is a global leader in food, environmental, pharmaceutical and cosmetic product testing and in agroscience CRO services. It is also one of the global independent market leaders in certain testing and laboratory services for genomics, discovery pharmacology, forensics, CDMO, advanced material sciences and in the support of clinical studies. In over just 30 years, Eurofins has grown from one laboratory in Nantes, France to 58,000 staff across a network of over 1,000 independent companies in 54 countries, operating 900 laboratories.



Principal Software Engineer (13740) - United States (Remote) - Remote Tech Jobs

#artificialintelligence

Getty Images works with over 496,000 contributors and image partners around the globe who add 8-10 million new assets each quarter to the 495 million assets contained in our catalogue. There are 2 fundamental questions the Getty Images Search & Ranking Team are working to solve: "If an image is worth a thousand words, wouldn't it be nice if we didn't have to type a thousand words to find it?" And: "What do we do when an image is your search query, instead of text entered into a field" To find a solution to these questions we are building a highly efficient, Artificially Intelligent (AI) Image search engine that pushes the boundaries of Natural Language Processing (NLP) and Visual Search Machine Learning (ML) The Search Team at Getty Images is responsible for building something that has never been built before. On a scale that will challenge the best of the best Engineers and Data Scientists alike. If you would like to be considered for an integral role on a high visibility team that has the deepest of impact on our industry and our customers, and you have expertise in some of the key skills and talents listed below, then we are very interested in speaking with you.


Remote NLP Engineer openings near you -Updated September 18, 2022 - Remote Tech Jobs

#artificialintelligence

Role requiring'No experience data provided' months of experience in None Architect data ingestion and text processing pipelines that will enable the development of useful tools and applications, including literature classification, chatbots, text summarization systems, and more. Qualifications: • Bachelor's Degree in computer science or related field and 4 years of experience; Masters Degree and 2 years experience • Advanced Python skills • Expertise in data structures and principles of optimal algorithm design • Experience working with large-scale data ingestion, both SQL (e.g., Hive, Impala) and NoSQL (e.g., Neo4J) • Some experience with NLP processing pipelines and/or text analysis • Interest in NLP and desire to learn about state-of-the-art NLP systems • Broad knowledge of AI/ML is a bonus, but is not required • Demonstrated ability to work with cloud infrastructure and tools (e.g., AWS, Cloudera) • Proficiency in a secondary programming language (e.g., Scala, Java) is preferred • Ability to multi-task and work within timelines Apply Here For Remote NLP Ops Engineer roles, visit Remote NLP Ops Engineer Roles Role requiring'No experience data provided' months of experience in New York Getty Images works with over 496,000 contributors and image partners around the globe who add 8-10 million new assets each quarter to the 495 million assets contained in our catalogue. There are 2 fundamental questions the Getty Images Search & Ranking Team are working to solve: "If an image is worth a thousand words, wouldn't it be nice if we didn't have to type a thousand words to find it?" And: "What do we do when an image is your search query, instead of text entered into a field" To find a solution to these questions we are building a highly efficient, Artificially Intelligent (AI) Image search engine that pushes the boundaries of Natural Language Processing (NLP) and Visual Search Machine Learning (ML) The Search Team at Getty Images is responsible for building something that has never been built before. On a scale that will challenge the best of the best Engineers and Data Scientists alike.


Reading Tea Leaves: Principles of Predictive Analytics and the Path to Time-Series Predictions

#artificialintelligence

Blog: medium @newfrontcreative Biography Scott Haines is a Principal Software Engineer on the Voice Insights team at Twilio. His focus has been on the architecture and development of a real-time (sub 250ms), highly available, trustworthy analytics system. His team is providing near real-time analytics that processes / aggregates and analyzes multiple terabytes of global sensor data daily. Scott helped drive Apache Spark adoption at Twilio and actively teaches and consulting teams internally. Scott's past experience was at Yahoo! where he built a real-time recommendation engine and targeted ranking / ratings analytics which helped serve personalized page content for millions of customers of Yahoo Games.