Skip to content
View kalpakprod's full-sized avatar
  • Kazakhstan
  • Joined Jul 21, 2026

Block or report kalpakprod

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 is supported. 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
kalpakprod/README.md

kalpakprod — routing, cost control, orchestration. Kazakhstan, remote.

An LLM router: ten or more providers behind one /v1 endpoint, each request scored on task fit, health, stability and remaining quota, then routed to the cheapest model that fits the role.

LLM infrastructure engineer. I keep a model router in production and spend most of my time on the boring half of that job: which provider is healthy right now, which model is cheap enough for this particular call, and how much of the prompt never needed to be sent.

The router

One OpenAI-compatible /v1 endpoint in front of ten or more providers. Every request is scored before it goes anywhere: how well a model fits the task, whether the provider is healthy, how stable it has been, and how much quota is left. The call goes to the cheapest model that clears the bar.

Cost comes down in four places, none of them clever:

  • Role-based routing. A planner gets the strongest model, an executor the cheapest one that can finish the job, a reviewer something in between. Most requests are executor requests.
  • Context compression before the call. The prompt is cut down before it reaches a provider, not after the invoice arrives.
  • Cache-hit stability. Prompts are kept in a shape that keeps hitting the prefix cache, which takes out both cost and first-token latency.
  • Per-provider rate limits with health-based fallback. A degraded provider drops out of scoring instead of failing requests.

Runs on Proxmox with Docker and LXC. Node and Python, state in SQLite, plain logs and cost metrics.

Public work

wheel — a Claude Code plugin that looks for an existing solution before the agent writes code. Five maturity flags, three gap axes, six adoption modes. MIT licensed. awesome-bc250 — a handbook taking the reader from a board in a box to running games on a PlayStation 5-derived APU. Sixteen sections, fifteen languages, distilled from 125 thousand community messages.

wheel scores candidate repositories on five deterministic maturity flags, measures three separate gaps between what a project gives and what you asked for, and returns an adoption mode instead of a diff. Writing from scratch is not one of the possible outcomes.

awesome-bc250 is the other half of the same habit: 125k messages of community chat compressed into a handbook that answers the questions in the order a person actually hits them, with the tuning trade-offs drawn as curves so nobody flashes a board before seeing what it costs.

Also

Multi-agent orchestration: sub-agent fan-out, reviewer and judge roles, swarm patterns. Python services and mitm proxies when an API needs to be understood rather than trusted. Networking and DevOps from the hardware up: Proxmox, WireGuard, mesh, VLESS.

Engineering degree. Based in Kazakhstan, working remotely.

Open to LLM engineering roles — @maxim_kukuruza on Telegram.

Popular repositories Loading

  1. wheel wheel Public

    The wheel already exists: a market-first gate for Claude Code that finds ready solutions before any line of code is written.

    Shell

  2. kalpakprod kalpakprod Public

    Profile

    JavaScript

  3. awesome-bc250 awesome-bc250 Public

    The newcomer's bible for the ASRock AMD BC-250 — a PS5-derived APU board turned into a budget Linux gaming & AI mini PC. 15 languages.

    Shell

  4. OmniRoute OmniRoute Public

    Forked from diegosouzapw/OmniRoute

    Never stop coding. Free MIT AI gateway: one endpoint, 339 providers (90+ free), 1200+ models — Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cl…

    TypeScript

  5. prime-agent prime-agent Public

    Forked from PrimeIntellect-ai/prime-agent

    A self-improving RLM agent for coding workflows and long-running autonomous tasks.

    TypeScript

  6. cross-impl-parity cross-impl-parity Public

    Оценочный контур для LLM-агента: вторая независимая реализация как источник истины вместо эталонного текста

    Python