Blog

Kairos: building a multimodal model with LFM2.5 and Kimi-K2.6

Kairos: building a multimodal model with LFM2.5 and Kimi-K2.6

Kairos (from the Greek καιρός): the opportune, fitting or decisive moment to do something. In ancient Greece it represented a qualitative time (...

AIMultimodalLoRA+5

Fredy Rivera

Fredy Rivera

Founder

Aug 19, 202625 min

TinyQwen: Understanding the Pipeline for Training an LLM from Scratch

TinyQwen: Understanding the Pipeline for Training an LLM from Scratch

Right now, some of the most talked about generative AI models are Large Language Models (LLMs), given the huge advances made in the field, going from ...

AIPre-TrainingFine-Tuning+4

Fredy Rivera

Fredy Rivera

Founder

Jul 26, 202635 min

Building Harvey: Legal AI Fine-Tuning for El Salvador

Building Harvey: Legal AI Fine-Tuning for El Salvador

How I built an AI model specialized in Salvadoran jurisprudence: from collecting 6,600+ judicial resolutions to generating synthetic data with legal reasoning traces to train an LLM with LoRA.

AIFine-TuningLoRA+7

Fredy Rivera

Fredy Rivera

Founder

Apr 7, 202614 min

Horizontal Scaling for Self-Hosted Image Generation

Horizontal Scaling for Self-Hosted Image Generation

How we overcame the vertical scaling barrier of AI models: when each new request doubled generation time, we migrated to batching + distributed inference. One forward pass, multiple images, automatic load balancing across GPUs. From bottleneck to scalable system.

AIImage GenerationFastAPI+3

Fredy Rivera

Fredy Rivera

Founder

Jan 12, 202610 min

Blog - Aquiles-ai