Machine learning

GGUF format and llama.cpp quantization for efficient CPU/GPU inference.

Evaluates LLMs across 60+ academic benchmarks (MMLU, HumanEval, GSM8K, TruthfulQA, HellaSwag).

Serverless GPU cloud platform for running ML workloads.

Parameter-efficient fine-tuning for LLMs using LoRA, QLoRA, and 25+ methods.

Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward…

Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching.

Guides migration, provisioning, search, log-analytics, trace-analytics, and Agentic AI Assistant workflows for Amazon OpenSearch Service and Serverless across…

Store and query vector embeddings using Amazon S3 Vectors, a cost-effective long-term vector storage service with its own API namespace (s3vectors).

Machine learning in Python with scikit-learn.

Model interpretability and explainability using SHAP (SHapley Additive exPlanations).

Team conventions for Python development — credentials, API clients, LLM response parsing, testing patterns, and data pipeline structure.

Run reproducible local or remote model evaluations with inspect-ai or LightEval, bounded smoke tests, captured configurations, and comparable results.

Build, test, and publish accessible Gradio interfaces for models and data workflows using reproducible Python environments.

Prepare and publish reproducible Hugging Face paper pages with linked code, models, datasets, demos, metadata, and validated claims.

Discover and analyze Hugging Face papers with traceable sources, linked models and datasets, and clearly separated claims and evidence.

Write research prompts that come back decision-ready — zero-prior-knowledge framing, evidence hierarchy, fact/inference separation, and a mandatory gap round.

Use for "interrogate", "adversarial review", "multi-model review", "challenge this", "stress test this code", "find blind spots", or "tear this apart".

Set up AI Runway on AKS — from bare cluster to running model.

Diagnose Day-2 AKS GPU and KAITO incidents using profile-aware, read-only evidence.

Debug live Azure Kubernetes Service (AKS) incidents with a read-only, evidence-first investigation.

Drizzle ORM conventions and best practices for TypeScript projects.

Search your own ADHX saves (posts kept from X, Instagram, TikTok, YouTube, Reddit and more) by meaning, including what is in pictures and videos, and fetch…

Research on the open web, in feeds and in notebooks: pick the method that fits the question — pulse for what people say right now, deep-research for a…

Audit the vault's PARA and Zettelkasten-inspired Atomic Note organization, linked knowledge, journals, metadata, LLM Wiki quality, and bounded statistics.