Vishal Veera Reddy
ReportIndian software engineer building open-source AI infrastructure. Sole maintainer of Lynkr, a self-hosted AI gateway that lets developers run any AI coding tool on any LLM provider.
Lynkr
A self-hosted AI gateway that decouples AI coding tools (Claude Code, Cursor, Codex, Cline, jcode, Pi) from their default LLM providers. Lynkr auto-detects the connecting tool, translates between Anthropic and OpenAI request formats, and routes to any of 12+ backends (Ollama, AWS Bedrock, Azure OpenAI, OpenRouter, Databricks, Moonshot, Google Vertex, llama.cpp, LM Studio, and more). A request-complexity classifier sends simple turns to free local models and complex ones to flagship cloud models, cutting per-developer AI bills 60-80% while removing vendor lock-in. Includes tool-result compression, MCP Code Mode (96% token reduction on tool definitions), persistent memory, and tier-based routing — all configured through a single .env file.
Core maintenance + roadmap (12 months)
Funds full-time work on Lynkr's core gateway: provider-format conversions, tier routing, tool-call translation across 10+ model formats (Minimax, Qwen, GLM, Llama, DeepSeek, Mistral), tool-result compression, persistent memory, MCP Code Mode, observability, tests, and docs. Estimated cost reflects one Indian maintainer working full-time for a year.
Infrastructure + benchmarks
Funds CI runners, benchmark harness for cost/quality/latency comparisons across providers, public dashboard at lynkr.dev, and self-hosted SearXNG + telemetry mirrors used by Lynkr's web search and routing layers.
Community sponsorship
Recurring small-amount sponsorship from individual developers and small teams who use Lynkr.
One-time contribution
Any-amount one-time contribution from users or supporters.