Architecting high-concurrency distributed backends in Go, autonomous AI agent pipelines, local-first offline architectures in Flutter, and vector search systems.
An autonomous AI assistant leveraging a distributed backend and an offline-first mobile app.
Concurrent fleet of 6 models (Groq Nemotron 120B, GPT-OSS 20B/120B, Qwen 27B, Gemini 3.5) with TPM/RPM quota control and automatic failover in Go.
Local microservice for embeddings and intent classification in <50ms.
Lexical search (SQLite v2 FTS5 BM25) + in-memory vector Cosine Similarity on the device.
Physical SQLite file isolation on client & PostgreSQL 16 with pgvector HNSW indices on the cloud.
Google Calendar OAuth2 integration with stealth offline resilience.