Maintenance-Eye

Overview Maintenance-Eye is a real-time AI copilot for physical infrastructure maintenance. Built for the Google Gemini Live Agent Challenge 2026, it lets a technician point a phone camera at equipment, speak naturally, and get grounded assistance without switching to a separate typing workflow. Links GitHub Repository Live Demo Status Maintenance-Eye is the shipped flagship in my public portfolio. The live deployment, repository, architecture diagram, infrastructure files, API routes, and tests are the evidence anchors for the case study. ...

Mar 1, 2026 · 3 min · Avishek Saha

GovtIntel

Status GovtIntel is in progress. It should be read as a current build, not as shipped product proof yet. The current repository has the structure for a federal procurement intelligence engine: ingestion modules, BM25 and vector retrieval scaffolding, prompt templates, FastAPI routes, typed models, and tests. The analysis endpoint is still a stub while the full RAG pipeline, demo, and evaluation suite are being completed. What It Is For Federal procurement data is useful but hard to scan quickly. The goal of GovtIntel is to turn contract records into structured intelligence briefs: contractor patterns, agency spending questions, market-entry signals, and evidence-backed answers over procurement data. ...

Apr 1, 2026 · 2 min · Avishek Saha