AI infrastructure engineer. Building production multi-agent systems on LangGraph, Gemini, and Redis — multi-tenant, multi-role, multi-timezone.
- Multi-agent AI platform — role-based LangGraph agents with long-term memory (Mem0 + Qdrant), Gemini context caching, and timezone-aware task scheduling. Runs in production across multiple organizations.
- Gemini context caching at scale — pre-baking tools + system instructions into
CachedContentper role per scenario, achieving a 85% token cost reduction on cached prompts across all agent turns.
| PR | Repo | What | Status |
|---|---|---|---|
| #1619 | langchain-ai/langchain-google |
Strip tools/tool_config/system_instruction from request when cached_content is set — production-tested fix for Gemini context caching with tool-calling agents |
🔍 In review |
| #161 | redis-developer/langgraph-redis |
Implement aprune() / prune() with keep_last=N strategy for interrupt-safe checkpoint pruning in AsyncRedisSaver and RedisSaver |
✅ Merged |
Core: Python · FastAPI · LangGraph · LangChain
AI: Gemini (context caching) · OpenAI (Whisper, TTS) · Mem0 · Qdrant
Infra: PostgreSQL · MongoDB · Redis · Celery · Docker · SQLAlchemy (async)
Tooling: uv · Alembic · Ruff · mypy · pytest
