Skip to content
View Arnav-Ajay's full-sized avatar
22 followers · 26 following

Achievements

Achievement: Arctic Code Vault Contributor

Achievements

Achievement: Arctic Code Vault Contributor

Block or report Arnav-Ajay

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Arnav-Ajay/README.md

👋 Hi, I’m Arnav

I’m focused on AI systems engineering, with an emphasis on understanding how LLM-based systems actually behave under real constraints.

My current work centers on:

  • Retrieval-Augmented Generation (RAG)
  • Agent architectures (planning, tools, memory)
  • Evaluation, observability, and failure modes in LLM systems

I’m learning by building small, inspectable systems and publishing them openly. Each repository is designed to isolate one idea, decision, or failure mode — starting from minimal baselines and adding complexity only when justified.

This is not about demos or frameworks.

Most repositories here are:

  • controlled experiments
  • system baselines
  • evaluation harnesses
  • learning artifacts with explicit assumptions and limits

The goal is long-term:
to develop sound intuitions about system design, tradeoffs, and failure — and to turn research ideas into working, testable code.

If you’re interested in how modern AI systems break — and how to reason about them, this profile is a public record of that process.

Pinned Loading

  1. rag-failure-modes rag-failure-modes Public

    Failure-first analysis of retrieval-augmented and agentic systems, focused on isolating and attributing failures across retrieval, planning, execution, memory, and policy layers.

    Python

  2. rag-systems-foundations rag-systems-foundations Public

    A systems-level analysis of static RAG pipelines, isolating ingestion, retrieval, and ranking boundaries to expose structural failure modes before generation.

  3. agent-systems-core agent-systems-core Public

    A systems-first implementation of agent control: explicit retrieval decisions, planner–executor separation, and auditable memory as core architectural mechanisms.

  4. agent-memory-systems agent-memory-systems Public

    A controlled, auditable implementation of agent memory that separates ephemeral state from persisted memory and exposes how policies govern state across runs.

    Python

  5. rag-reranking-playground rag-reranking-playground Public

    Controlled experiment isolating reranking as a first-class RAG system boundary, measuring how evidence priority—not recall—changes retrieval outcomes.

    Python

  6. rag-retrieval-eval-harness rag-retrieval-eval-harness Public

    Minimal harness for evaluating retrieval behavior in RAG systems

    Python