Rivo Careers
Co-Founder
**Role:** Co-Founder **Period:** November 2025 to present **Stack:** Next.js, TypeScript, Tailwind, Supabase, Python, FastAPI, WebRTC
What it is
Rivo Careers is a career-tech platform for Canadian students and new grads. It covers six product domains: AI mock interviews, ML job matching, automated job aggregation, resume tools, application tracking, and career resources. I built the engineering and own the financial model.
The engineering
64 API routes across 6 domains
The platform runs on Next.js with a Supabase backend. The API surface covers authentication, job listings, user profiles, interview sessions, matching results, and admin tooling. Each domain has its own route group, middleware, and validation layer.
AI mock interview system
The interview system pairs a lip-synced humanoid avatar with live speech transcription. The candidate speaks into their mic; the system transcribes in real time, runs the response through an evaluation pipeline, and scores across three categories: behavioral, technical, and communication. The avatar responds with follow-up questions calibrated to the role.
The pipeline: WebRTC for media capture, speech-to-text for transcription, LLM for evaluation and follow-up generation, text-to-speech for avatar audio, and a lip-sync engine that maps phonemes to blend shapes on the avatar mesh.
Two-phase ML job matching engine
Phase one: NLP-based resume scoring. The system embeds the candidate's resume using TF-IDF vectors, then computes cosine similarity against job description embeddings. This produces a raw relevance score for every listing.
Phase two: behavioral signal blending. The system tracks which jobs the candidate views, saves, applies to, and dismisses. These signals feed a secondary scoring model that adjusts the NLP scores based on revealed preferences. The blend ratio is configurable per user via a strictness slider.
Automated job aggregation pipeline
A daily pipeline sources internship listings from across the Canadian market, deduplicates against existing records, validates listing metadata (company, location, deadline, requirements), and flags expired or suspicious entries through a multi-layer expiry detection system. The goal: thousands of validated listings, refreshed daily, with minimal manual oversight.
The business
I built the financial models that project unit economics, CAC/LTV ratios, and runway scenarios. These informed our pricing strategy and go-to-market decisions. I conducted market sizing across the North American career services landscape (multi-billion dollar TAM) and built the competitive positioning that differentiates us from incumbents.
I present growth metrics, KPI performance, and strategic roadmaps to advisors and prospective investors.