Path 02 · full reference map
AI Systems & Inference Engineering
The complete path from ML foundations and transformer internals to efficient, observable, distributed AI infrastructure.
Prior practice · disclosed carefully
I am not starting this path at zero.
These claims come from résumé-backed work. They are intentionally generalized, so the site shows what I have applied without inventing public artifacts or exposing private systems.
Agent intelligence
Built an AI classifier that turns trace behavior into structured insights and routes work between models and deterministic code.
Evaluation systems
Built failure taxonomies, automated evaluation workflows, regression detection, and quality monitoring with RAGAS and Langfuse.
Production optimization
Applied routing decisions to reduce unnecessary model calls and improve cost and latency at the orchestration layer.
Published progress
8%6 of 80 topics complete · updated with each site release
Quarter 01
Machine-learning foundations
Build enough mathematical and implementation fluency to reason about model behavior.
Quarter 01
Transformer internals
Follow the model from tokens through attention and generation.
Quarter 01
Retrieval systems
Build retrieval as an evaluated information system, not a demo feature.
Quarter 02
LLM applications and agents
Design controllable application loops around probabilistic models.
Quarter 02
Evaluation
Make quality measurable across prompts, models, retrieval, and agents.
Quarter 03
Inference engineering
Understand latency, throughput, memory, and scheduling at serving time.
Quarter 03
GPU and kernel foundations
Connect serving behavior to memory movement and compute.
Quarter 03
Distributed training
Study scale, faults, and communication across training clusters.
Quarter 04
Production AI platform
Operate AI as a reliable multi-tenant system.
Quarter 04
Systems builds
Consolidate the path in complete engineering artifacts.