Skip to content
View YusefSyed's full-sized avatar

Highlights

  • Pro

Block or report YusefSyed

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
YusefSyed/README.md

Yusef Syed

University of Toronto student pursuing Mathematics & Statistics, building AI evaluation systems and reliable software products.

I work across Python evaluation infrastructure, TypeScript/React Native products, and backend reliability. I care about reproducible evidence, failure modes, and honest limits—not just demo paths.

Product portfolio · Résumé (PDF) · Email

Internship availability

Term Availability
Summer 2027 Available full-time from May through August for software engineering, AI product, evaluation/reliability, backend/platform, or mobile roles
Fall / Winter Open to selective AI/software internships that are compatible with University of Toronto coursework
Location Canada or the United States; open to relocation and remote work
Work authorization U.S.–Canadian dual citizen; no sponsorship required in either country

Expected graduation: May 2030.

Selected work

Agent Eval Mutation Lab · Python, Inspect, Docker, SQLite

A typed, deterministic engine for testing execution-semantic robustness in tool-agent scorers. It runs 104 canonical tasks with resumable SQLite state, content-addressed evidence, explicit unknown/abstain handling, and clean-checkout artifact reproduction. Its public reports retain failed model-study gates and unfavorable outcomes instead of promoting unsupported conclusions.

Shipped products

Project What I built Evidence
Aesthetics AI Released iOS fitness product using React Native, Expo, Supabase/Postgres, structured model outputs, and RevenueCat. Documented reliability work covers persist-before-sync workout completion, atomic writes, restart recovery, and account isolation. App Store · Case study
Tiraz Released privacy-first iOS wardrobe app plus a public PyTorch garment-completion study. Three seeded models reached 58.5–59.0% held-out top-1 accuracy versus a 52.7% co-occurrence baseline; the missing-context coverage failure remains visible. App Store · ML study · Case study
CallReclaim Private missed-call recovery MVP with consent-aware SMS, validated AI lead extraction, signed provider webhooks, owner-scoped data, and PostgreSQL reconciliation for uncertain send outcomes. No production traffic or revenue is claimed. Sample-data demo · Case study

Developer tools and open source

Technical toolkit

  • Languages: Python, TypeScript, JavaScript, SQL
  • AI and evaluation: PyTorch, NumPy, Inspect, structured LLM APIs, deterministic evaluation pipelines
  • Web and mobile: React, React Native, Expo, Next.js, Node.js
  • Data and reliability: PostgreSQL, Supabase, SQLite, Docker, pytest, mypy, GitHub Actions
  • Product integrations: Twilio Voice/SMS, RevenueCat, Vercel

How I build

  • Turn ambiguous product requirements into tested systems with explicit failure behavior.
  • Treat dependencies, logs, provider callbacks, and external inputs as untrusted evidence.
  • Separate local or synthetic verification from production, user, revenue, and impact claims.

If you are hiring for a Summer 2027 or selective Fall/Winter AI/software internship, email me.

Pinned Loading

  1. agent-eval-mutation-lab agent-eval-mutation-lab Public

    Typed deterministic Python engine for testing execution-semantic robustness in tool-agent scorers.

    Python

  2. tiraz-garment-completion tiraz-garment-completion Public

    Annotation-only PyTorch garment completion with source-group splits, calibration, and missing-context evaluation

    Python

  3. agent-proof agent-proof Public

    TypeScript CLI for running approved checks and recording reproducible verification evidence

    TypeScript