Chirp Labs
AI sales-intelligence platform: a unified synthetic data layer, multi-agent workflows, and streaming deal-intelligence UI built with Next.js.
Full Stack AI Software Engineer
2025–2026
Unified Synthetic Data Layer
Sales context lives in many places: emails, call transcripts, CRM events, and calendars. I rebuilt the ingestion pipeline so all of these flow through background jobs and external connectors into a single data layer, then used LLMs to distil raw sales conversations into structured, queryable insights.
Low-Latency Semantic Retrieval
On top of the raw data I designed synthetic intelligence layers that capture nuanced sales signals and make them retrievable semantically. This gave the product faster, more accurate answers about the state of each deal without re-reading every underlying conversation.
Streaming Chat & Deal Intelligence UI
I built the streaming chat and deal-intelligence interfaces with Next.js, React, TanStack Query, and Zustand. Synthesised insights and next-best actions stream into the UI as they are generated, so users see results immediately rather than waiting for a full response.
“Chirp was about turning messy, unstructured sales conversations into something a team can actually act on — fast enough that it feels like part of the product rather than a report.”