Skip to content
View anirket's full-sized avatar
:octocat:
:octocat:
28 followers · 29 following

Achievements

Achievement: Pair Extraordinairex3Achievement: QuickdrawAchievement: Pull Sharkx3Achievement: YOLO

Achievements

Achievement: Pair Extraordinairex3Achievement: QuickdrawAchievement: Pull Sharkx3Achievement: YOLO

Organizations

@smallcase

Block or report anirket

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.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

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

Report abuse
anirket/README.md

Hey 👋 I'm Aniket

Learning to build with LLMs — agents, evals, retrieval, and the glue that holds them together.

Mostly Python these days. Working through tool-calling loops, structured outputs, RAG, MCP and how to actually measure whether any of it works, one small project at a time.

Four years shipping production software before this.

Pinned Loading

  1. ask-my-filings ask-my-filings Public

    RAG over Indian company filings with measured retrieval: hybrid search, cited answers or honest refusals, and a head-to-head against long-context.

    Python

  2. mini-agent mini-agent Public

    An AI agent loop built by hand with the Anthropic SDK (no framework), then rebuilt in Pydantic AI to compare. 3 tools, error handling, step/token limits, summarising old turns, and a 20-task eval: …

    Python

  3. smallest-eval smallest-eval Public

    Evals in CI for an LLM app: promptfoo runs code checks, an LLM judge and cost/latency limits on every PR, and blocks the merge if answers get worse. The judge is checked against hand labels (κ = 1.…

    JavaScript

  4. jd-lens jd-lens Public

    Paste a job description, get strict JSON (role, seniority, must-haves, stack), a fit score against your resume, and tailored resume bullets. Every match must quote real resume text, which is checke…

    Python