Abhijeet Kadam
About
I came into software sideways. An Electronics and Telecommunications degree left me more curious about the code running on the hardware than the hardware itself, so I taught myself the rest by building things instead of waiting for a course to explain them.
Skills
Experience
AEOS Labs
Full Stack AI Engineer
- Built a full-stack B2B procurement platform from the ground up. Owned the backend infrastructure and a Next.js frontend (Shadcn/ui, Redux Toolkit, TanStack Query), shipping marketplace, RFQ, and vendor management workflows from UX designs to production.
- Cut RFQ quote turnaround from 48+ hours to minutes. An AI-powered pipeline parses inbound vendor emails and document attachments (PDF, DOCX, Excel) straight into structured quote line items.
- Architected a TypeScript monorepo with two independently deployable services, an HTTP API server and an async background job worker, and engineered a bidirectional sync with Microsoft Business Central covering vendors, items, quotes, and conversation threads, with real-time email ingestion via Microsoft Graph webhooks.
- Made the system resilient by default: circuit breakers on every external call, exponential backoff with dead-letter queues, cluster-aware Redis rate limiting, and a multi-strategy vendor sourcing orchestrator that cascades through catalog matching, preferred vendor lookup, AI-powered web discovery, and manual review fallback.
- Deployed the full stack on Azure, Container Apps with independent per-service scaling, Azure PostgreSQL, Azure Redis, Blob Storage for documents, and Azure Communication Services for transactional and follow-up emails.
Work
Rika
Joins your Zoom, Google Meet, and Teams calls. Captures speaker-attributed transcripts and answers questions with cited RAG, live in-call through @Rika mentions or after the meeting in a web chat.
ARIA
Answers inbound calls end to end. Auto-detects the caller's language across 30+ languages, books appointments, and handles real-time barge-in at sub-800ms latency through a native audio-to-audio pipeline.
redai
A provider-agnostic TypeScript agent SDK published on npm. Hand-written agent loop, tool calling, structured output, guardrails, multi-agent handoffs, sessions, streaming, tracing. No agent framework underneath.
NotebookLM
Feed it PDFs, DOCX, websites, YouTube videos, transcripts. Ask it anything in plain language and get a streamed answer with inline citations that open the exact source passage.