Research And Development Specialist

R&D Engineer — LLMs & Rust — Exvisit

Compensation: Equity

About Exvisit Exvisit is a hybrid cloud agentic IDE. A Rust daemon runs locally (file I/O, git, bash, symbol-graph queries) while a cloud engine runs the agent loop and model calls over an encrypted connection — files and credentials stay on the user's machine. Every AI-proposed edit is verified locally — shadow compiler, AST masking, symbol-graph blast-radius scoring — before it touches disk.

The Role You start hands-on, not sidelined: build mini-projects that mirror Exvisit's own model — a small Rust daemon, a tool-calling agent loop, a toy verification layer — the same problems we solve in production, at a smaller scale. Weekly discussions double as demo day — show what you built, get direct feedback. As you find your footing, you move into the real daemon, the real agent loop, and open research problems. Alongside the engineering, you sit in with the founding team on business, outreach, and distribution — this is a seat in the whole Exvisit journey, not just your corner of it.

What You'll Work On

• Agent loop — ReAct-style tool-calling loop; debug failures, build self-correction
• Provider adapters — normalize tool-calling/streaming differences across model APIs
• Rust daemon — workspace tools, PTY shell, git, symbol-graph queries
• Structural verification — shadow compiler, AST masking rules, blast-radius scoring
• Context management — token budgeting, auto-compaction, context-window discovery
• Extensibility — MCP client layer + Skills system
• Cloud/local protocol — websocket contract between daemon and cloud brain

What We're Looking For

• Strong Rust — async, ownership, systems software
• Real hands-on LLM experience — prompting, tool-calling, agent loops (not just calling a chat API)
• Comfortable debugging distributed/websocket systems
• Bias toward verification over "it looked fine when I tried it"
• Sharp, hypothesis-driven debugging instincts

Nice to Have

• Tree-sitter, ASTs, or code-intelligence protocols (SCIP, LSIF, LSP)
• Tauri or other Rust desktop app experience
• OSS contributions — Rust, LLM tooling, or dev-tools infra
• Embeddings/vector search, or local model inference (e.g. llama.cpp)

How to Apply Send your resume, GitHub, or relevant projects to

hello@exvisit.in . A project or writeup on correctness or LLM tool-use beats a polished resume.