Data Scientist with 8+ years building LLM-powered applications, computer vision systems, and NLP pipelines — with a focus on financial compliance and responsible AI. Writing deep-dives on Generative AI, Model Risk Management, and where regulation meets technology.
Something that comes up a lot in financial services AI: the gap between running an LLM eval and producing something regulators can actually work with. ROUGE scores don't tell you which SR 11-7 clause you're satisfying. B…
𝑪𝒍𝒂𝒖𝒅𝒆 𝒋𝒖𝒔𝒕 𝒑𝒓𝒐𝒗𝒆𝒅 𝒚𝒐𝒖𝒓 𝒎𝒐𝒅𝒆𝒍 𝒗𝒂𝒍𝒊𝒅𝒂𝒕𝒊𝒐𝒏 𝒎𝒆𝒕𝒓𝒊𝒄𝒔 𝒎𝒊𝒈𝒉𝒕 𝒃𝒆 𝒂 𝒍𝒊𝒆. Anthropic’s recent report on "Eval Awareness" is more than just a cool AI story. For Model Risk Manage…
𝑾𝒆 𝒏𝒆𝒆𝒅 𝒕𝒐 𝒔𝒕𝒐𝒑 𝒕𝒓𝒆𝒂𝒕𝒊𝒏𝒈 𝑳𝑳𝑴𝒔 𝒍𝒊𝒌𝒆 𝒕𝒉𝒆𝒚 𝒔𝒑𝒆𝒂𝒌 𝑬𝒏𝒈𝒍𝒊𝒔𝒉. 🤐 Following up on that "Prompt Repetition" paper from Google (arXiv:2512.14982), a few other findings are making it clea…
The Federal Reserve's November 2025 Statement of Supervisory Operating Principles signals a seismic shift — from checkbox compliance to material risk. Here's what changed and why it matters.
Fast.ai uncovered something strange in LLM fine-tuning: training loss dropped suddenly after just one pass through the data — suggesting models can memorize inputs almost immediately. Here's what it means.
A comprehensive framework for analyzing open-source GenAI across near, mid, and long-term development stages — and why the benefits generally outweigh the risks when governance keeps pace.
Re-training LLMs from scratch when new data arrives is prohibitively expensive. Three simple strategies — LR re-warming, LR re-decaying, and minimal data replay — match the performance of full re-training at a fraction of the cost.
Traditional ensemble methods fail when correct answers are in the minority. AoR introduces hierarchical reasoning chain evaluation and dynamic sampling to fix this — and consistently outperforms standard approaches.
