Mahyar Jaberi

View my work Get in touch

Builds: AI agent systems and full-stack products
Now: AI Solution Engineer co-op at FGF Brands
Studying: Honours BA, Computer Science (Co-op), York University
Based: Aurora, ON

Mahyar standing on top of Preikestolen with his arms open, above the fjord in Norway

Mahyar playing tennis on an outdoor court

Mahyar taking a selfie with his team at the OpenAI hackathon

Mahyar talking with his team at the TMLS hackathon

Mahyar standing on a wooden dock by the water in Norway, with mountains behind him

About

A bit about me

Travel

I love to travel. This is me on top of Preikestolen, in Norway.

Play

I really like playing sports, like football and tennis.

Train

I run and lift weights.

Build

And I build random stuff, like at the OpenAI and TMLS hackathons.

My journey started in Grade 11, when I was forced to take computer science, and at the time I hated it. But it gave me a new way of thinking, and my creativity slowly moved from drawing imaginary creatures on paper to turning my ideas into code. Now here I am, always building something to keep my inner child happy. Thanks for reading.

Selected work

Projects

Live in stores

BowlWise

Scores 260 dog foods against your dog

Coming soon

MoneyMind

An AI money coach

The shortlist in three states: four Toronto listings reordered after the renter says parking matters most, all four on the phone at once, then rewritten from the calls as booked, no answer, over budget and leased

Built in a day

Realest

Calls listing agents so renters don’t have to

Built in 5 days

Maridian

Six agents triaging defects on a bakery line

In progress

Graph-Native Retrieval for Clinical LLM Agents

Directed studies, York Data Mining Lab

Barcelona’s passing network from the 5–0 Clásico: eleven player nodes joined by arrows, with Xavi the largest node, at the centre

Research

Tactical DNA

Does a coach leave a fingerprint on passing?

Experience

Where I’ve worked

FG
AI Solution Engineer Co-op
FGF Brands · Toronto, ON (Hybrid)
NowSept 2026 – Present
A one-year co-op term building AI agent systems.
- Building a promotion pipeline that moves AI agents between environments without a rebuild, to cut deployment overhead and configuration drift.
- Adding live web search to AI chat agents: the backend services, the frontend integration, source citations, and the testing and validation tooling.
- Working in Python, Azure services and Docker with containerized deployments, and writing the technical docs and architecture guides for onboarding.
- Python
- FastAPI
- React
- Azure
- Docker
YU
Directed Studies, EECS 4070
York Data Mining Lab · Prof. Manos Papagelis
Fall 2026 · Toronto, ON
Research on retrieval for LLM agents that answer clinical questions over FHIR health records.
- Building a graph traversal tool that lets clinical LLM agents follow the links between a patient’s FHIR records, instead of fetching each linked record themselves.
- Setting up FHIR‑AgentBench on the MIMIC‑IV FHIR demo: 928,935 records of 13 types, checksum-verified, including every ground-truth record behind the benchmark’s 2,931 questions.
- Comparing the same agent with and without the tool; for reference, the benchmark paper’s best agent reaches 50% answer correctness (Lee et al., ML4H 2025).
- Python
- FHIR
- LiteLLM
- pandas
PV
Assistant Manager
Pet Valu · Richmond Hill, ON
Aug 2023 – Aug 2026
1. Assistant Manager, from Aug 2025
2. Retail Sales Associate, from Aug 2023
Three years on the retail floor, promoted to Assistant Manager, and the reason BowlWise exists.
- Trained new staff on service, nutrition basics, store routines, safety and the POS, and helped plan schedules and tasks.
- Tracked weekly sales patterns and adjusted product placement, signage and recommendations to match.
- Helped customers choose food by breed size, age, allergies and dietary goals.
NV
Software Developer Intern
Nova Ventures · Toronto, ON (Hybrid)
May – Aug 2025
A goalkeeper analytics platform for competitive soccer teams.
- Built backend API endpoints in Python and FastAPI for a goalkeeper analytics platform, exposing per-session performance data to the coaching dashboard.
- Built a data pipeline that parsed and cleaned match stats from club PDFs and spreadsheets into structured database tables for analysis.
- Built interactive React heatmap components that visualized save patterns across goal zones.
- Python
- FastAPI
- React
- TypeScript
ST
Software Engineer Intern
Sepantech · Sweden (Remote)
Mar – Aug 2024
A course-management platform for healthcare and physiotherapy professionals.
- Built core features for a course-management platform used by 500+ healthcare and physiotherapy professionals across Sweden: enrollment, progress tracking and automated certification.
- Designed 6+ MySQL tables in a normalized schema and optimized 12+ queries with composite indexes, cutting average response time by about 30%.
- Built 8+ Node.js API endpoints for authentication, enrollment, progress updates and certificate generation, in an Agile team.
- Node.js
- MySQL