AI & ML

Evaluates LLMs across 100+ benchmarks from 18+ harnesses (MMLU, HumanEval, GSM8K, safety, VLM) with multi-backend execution on local Docker, Slurm HPC, or cloud platforms.

Generate high-quality sentence and text embeddings for semantic similarity, clustering, and retrieval using 5000+ pre-trained models. Supports multilingual and domain-specific embeddings for RAG and semantic search.

Train and experiment with a minimal GPT implementation in ~300 lines of PyTorch, from character-level Shakespeare to GPT-2 scale.

Manage and use Lambda Labs GPU cloud instances for ML training and inference with SSH access, persistent filesystems, and multi-node clusters.

Deploy Meta's LlamaGuard moderation model to filter LLM inputs and outputs across 6 safety categories using HuggingFace, vLLM, or FastAPI.

Run LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware. Use for edge deployment, M1/M2/M3 Macs, AMD/Intel GPUs, or when CUDA is unavailable. Supports GGUF quantization (1.5-8 bit) for reduced memory and 4-10× speedup vs PyTorch on CPU.

Control LLM output with regex and grammars to guarantee valid JSON, XML, or code generation, enforce structured formats, and build multi-step workflows using Microsoft Research's Guidance framework.

Segment any object in images using points, boxes, or masks as prompts, or automatically generate all object masks with zero-shot transfer.

Generate images from text prompts, perform image-to-image translation, inpainting, and build custom diffusion pipelines using Stable Diffusion models via HuggingFace Diffusers.

GPU-accelerated data curation for LLM training, supporting text, image, video, and audio with fuzzy deduplication, quality filtering, semantic deduplication, PII redaction, and NSFW detection.

Optimizes transformer attention with Flash Attention for 2-4x speedup and 10-20x memory reduction. Supports PyTorch native SDPA, flash-attn library, H100 FP8, and sliding window attention.

Fine-tune and align language models using reinforcement learning with TRL, including SFT, DPO, PPO, GRPO, and reward model training.

Detect prompt injections and jailbreak attempts in LLM applications using Meta's 86M parameter classifier. Filter user inputs, third-party data, and RAG documents with low latency and multilingual support.

Extract structured data from LLM responses with Pydantic validation, automatic retries, and streaming support across multiple providers.

Expert guidance for implementing GRPO/RL fine-tuning with TRL for reasoning and task-specific model training.

Optimizes LLM inference with NVIDIA TensorRT for maximum throughput and lowest latency on NVIDIA GPUs (A100/H100).

Add distributed training support to any PyTorch script with minimal code changes using a unified API for DDP, DeepSpeed, FSDP, and mixed precision.

Add programmable safety guardrails to LLM applications at runtime, including jailbreak detection, input/output validation, fact-checking, hallucination detection, PII filtering, and toxicity detection.

Extend context windows of transformer models using RoPE, YaRN, ALiBi, and position interpolation techniques for processing long documents and implementing efficient positional encodings.

Train Mixture of Experts (MoE) models using DeepSpeed or HuggingFace, covering architectures, routing, load balancing, and expert parallelism.

Merge multiple fine-tuned models using mergekit to combine capabilities without retraining, covering SLERP, TIES-Merging, DARE, Task Arithmetic, linear merging, and production deployment strategies.

Train AI models to be harmless through self-critique and AI feedback using a set of constitutional principles, without requiring human labels for harmful outputs.

Trains large language models (2B-462B parameters) using NVIDIA Megatron-Core with advanced parallelism strategies for maximum GPU efficiency.

Adds PyTorch FSDP2 (fully_shard) to training scripts with correct init, sharding, mixed precision/offload config, and distributed checkpointing. Use when models exceed single-GPU memory or when you need DTensor-based sharding with DeviceMesh.

Perform causal interventions on PyTorch models using pyvene's declarative framework for causal tracing, activation patching, and interchange intervention training.

Run interpretability experiments on neural network internals using nnsight, with optional NDIF remote execution for massive models.

Train and analyze Sparse Autoencoders (SAEs) using SAELens to decompose neural network activations into interpretable features for mechanistic interpretability research.

Compiles research inputs—PDFs, code, logs, notes—into structured Agent-Native Research Artifacts with cognitive and physical layers.

Accelerate LLM inference using speculative decoding, Medusa multiple heads, and lookahead decoding techniques for 1.5-3.6× speedup without quality loss.

Inspect and manipulate transformer internals via HookPoints and activation caching for mechanistic interpretability research.

Performs a semantic epistemic review of Agent-Native Research Artifacts, scoring six dimensions and producing a constructive report with a recommendation.

Train a personalized identity model on a person's face for use in Higgsfield's image and video generation.

Generate images, videos, 3D assets, and audio via the Higgsfield AI CLI, including Marketing Studio ads and Virality Predictor analysis.

Install and configure Acontext, a memory layer for AI agents that provides persistent sessions, file storage, and skill management.

Write Python code in n8n Code nodes, covering data access patterns, return formats, and standard library limitations.

Handle files and binary data in n8n workflows correctly, covering the $binary vs $json split, reading/writing binary, preserving binary across transforms, and the agent-tool binary boundary.