Amirali Tolooei.
Full-Stack & AI Developer, Toronto
Computer Science, University of Toronto. Currently building voice agents and LLM evaluation tooling at HouseSigma. Interested in full-stack, ML, and games — in roughly that order, on a given Tuesday.
About Me
My name is Amirali Tolooei, a computer science major and statistics minor graduate from the University of Toronto. In school I spent a lot of time learning Machine Learning and Human-Computer Interaction. Since graduating, I've shipped AI and full-stack work at three companies — voice cloning at Montagy, fleet finance at Cypress, and now applied AI at HouseSigma — and used the gaps between to chase whatever caught my attention.
Every corner of computer science feels worth the detour.
I've been a huge video game nerd since I was a kid, and that's really what ignited my passion for computer science to begin with. During my studies AI hit its huge boom and I ended up making it my focus, doing a concentration in AI and Human-Computer Interaction. Lately that curiosity has spilled in every direction: built this site, set up a homelab server, trained SqueeBot (a Discord chatbot styled as a Magic: The Gathering character, via QLoRA fine-tuning), and replaced my OS with Arch Linux to dig into how operating systems actually work (I use Arch, btw).
Projects
TableTally
A deck-building companion for Magic: The Gathering players that searches friends' Moxfield collections and local game store inventories.
SqueeBot
A Discord chatbot styled as a Magic: The Gathering character, with a three-tier LLM cascade, local RAG over 10K embeddings, per-user memory, and a self-hosted QLoRA fine-tune of Llama 3.1 8B.
BookShare
A book-sharing platform for friend groups to browse shared bookshelves, track ownership, leave ratings and reviews, and discover what each other are reading.
FinTrack
A personal finance dashboard that ingests bank statement PDFs, tracks balances across accounts, and calculates net worth over time.
Size Matters
A single-player puzzle experience, where players are challenged to cross a series of rooms by completing puzzles that can only be solved using the Swap Mechanic.
Experience
HouseSigma
Full-Stack Developer
Jun 2026 — Present
· Remote — Toronto, ON
Building applied-AI systems for a Canadian real-estate platform: a real-time voice agent that runs candidate pre-screens, and an evaluation platform for vision-language models in the listing photo pipeline.
- •Built and deployed an AI voice agent that conducts real-time phone pre-screen interviews, using a Twilio + Deepgram speech pipeline with concurrent LLM classifiers for turn-taking and adaptive follow-ups
- •Built an LLM-persona test harness that exercises the live agent with synthetic candidates graded on a multi-axis rubric, and designed the real-time audio media bridge the calling system runs on
- •Designed and built an end-to-end LLM evaluation platform for a real-estate photo pipeline, benchmarking 15+ vision-language models against a human-verified golden set with a validated LLM-as-judge cascade
- •Improved model output quality through iterative prompt and context engineering, and produced inference cost analysis to inform production-scale deployment
- Python
- Twilio
- Deepgram
- AWS Bedrock
- S3
- LLMs/VLMs
- Kafka
- GitHub Actions
Cypress
Founding Engineer & Full-Stack Developer
Jan 2026 — Apr 2026
· Remote — San Francisco, CA
Led full-stack development of an MVP fleet finance platform from the ground up for a startup in the automotive remarketing space, working alongside a senior developer and designer. Responsible for backend architecture, frontend implementation, cloud infrastructure, and CI/CD.
- •Designed and built a Django REST Framework API with JWT authentication, 3-tier role-based access control, invitation-based onboarding, and a configurable pricing policy engine integrating Cox Automotive's MMR API
- •Built a React 18 + TypeScript SPA with shadcn/ui, implementing a Figma-to-code design system with role-conditional dashboards, data tables, and a multi-step signup wizard
- •Provisioned staging infrastructure on GCP — Cloud Run, Cloud SQL (PostgreSQL 16), and Artifact Registry — with multi-stage Docker images and cross-platform support
- •Enforced strict code quality with mypy strict mode, ruff linting, AST-based schema validation, and 400+ automated tests
- •Helped grow the company from first revenue to over $12M in annualized GMV in its first quarter of trading
- Python
- Django
- React
- TypeScript
- Tailwind CSS
- shadcn/ui
- PostgreSQL
- GCP
- Cloud Run
- Docker
Montagy AI
AI Developer
May 2025 — Oct 2025
· Toronto, ON
Contributed to a voice cloning and text-to-speech platform focused on generating personalized speech synthesis from user voice recordings, with multilingual capabilities and emotion blending.
- •Deployed a zero-shot TTS system using a diffusion transformer model in a Python Flask application that records user voices and generates personalized speech synthesis
- •Optimized production model pipelines with FastAPI, implementing server-level caching strategies to achieve 98% uptime and reduce latency by 40%
- •Containerized applications with Docker and established CI/CD pipelines with GitHub Actions, automating testing, linting, and deployment workflows
- Python
- Flask
- FastAPI
- SQLite
- ChromaDB
- AWS
- Docker
- GitHub Actions
Contact
Feel free to check out my Github or LinkedIn, or shoot me an email if you want to chat!