Software engineer at TD Bank, building agentic automation.
I'm Mondher Ilyes Izemrane. I build production software across AI, full-stack systems, and automation—most recently at TD Bank and on the Ontario Line with SYSTRA Canada.
Experience
Production engineering across banking, transit, hospitality analytics, and career technology.
- Built a human-in-the-loop agentic automation that reduced onboarding support load and service-request preparation time, reaching finalist status in an internal innovation challenge.
- Cut test-case generation time by 50%+ across 7+ legacy processes while supporting migration to a new RPA application.
- Engineered source-artifact inconsistency detection to accelerate subject-matter expert and business analyst review.
- Designed and deployed a production LLM assurance platform that reduced manual review across 800+ safety requirements.
- Kept token usage to 60% of the allocated budget while processing CAD-heavy evidence packages exceeding 1,000 pages per requirement.
- Enabled consistent review of thousands of designer-submitted artifacts across multiple contracts and seven Ontario Line South Civil stations.
- Built an analytics platform that reduced hospitality demand-forecasting time by 80%.
- Implemented a reusable AI agent scraper to collect, structure, and validate event data from five sources.
- Introduced real-time anomaly alerts that removed the need for manual demand analysis.
- Deployed Python endpoints that automated career coaching for 200+ users, saving more than five hours per user each week.
- Built a fault-tolerant email automation system with Celery and Redis for onboarding and retention flows.
- Improved platform load capacity by 30% while preventing user-facing crashes before launch.
Earlier experience and leadership
Work that shaped how I communicate technical ideas, lead rooms, and turn early concepts into usable systems.
DeepPixel
Machine Learning Consultant
Developed an educational generative-AI product for more than 10,000 children, cut infrastructure and inference costs by 90%, and translated stakeholder needs into an implementation-ready technical proposal.
University of Toronto
Computer Science Transition Coach
Mentored 12+ first-year computer science students through weekly academic strategy sessions and used program feedback to strengthen retention-focused coaching materials.
BACSA
Technology Director
Designed BioPython and BLAST workshops for 50+ students and coordinated the technical delivery of bioinformatics events, panels, and coding initiatives.
Selected system
One independent product, built across the entire stack.
FightSight
Independent full-stack engineer · 2024—NowA combat-sports prediction platform engineered from raw fighter history to authenticated, production-grade decisions.
Visit FightSight.devProducts built beyond the prototype.
Independent systems shaped across product strategy, architecture, data, interface design, and production delivery.
A mobile wellness system that turns wearable data into clear, explainable signals across readiness, recovery, strain, sleep, and long-term health.
- Built an Expo and Fastify platform that ingests real Google Health data through backend-managed OAuth, normalizes wearable metrics, and recomputes user scores.
- Designed deterministic readiness, recovery, strain, stress-pattern, and experimental healthspan models using chronological backfills and a 42-day signal window.
- Created interactive 7, 30, and 90-day trend views with scrubbing, baselines, averages, and signed changes so users can understand patterns—not just daily numbers.
A three-sided marketplace designed for how clients, service providers, and operations teams actually coordinate home services in Morocco.
- Engineered a role-gated Expo application covering phone onboarding, service discovery, scheduled bookings, provider offers, KYC gates, and earnings workflows.
- Built marketplace services with Fastify, Prisma, PostgreSQL, Redis, and BullMQ, including first-accept-wins matching and operations tooling.
- Designed for Moroccan use with Darija, French, and Arabic localization, RTL support, consent enforcement, privacy requests, location permissions, and reduced-motion-aware interactions.
More products and client work
Earlier builds spanning collaborative media, generative AI, and client-led web development.
Loggit
Founder · Product design and engineeringDesigned a private event photo-sharing product that replaces scattered group chats and cloud folders with organized feeds, collaborative uploads, and a trust-first experience.
Independent productEngineering breadth, grounded in outcomes.
I work best where the system is ambiguous, the evidence is messy, and reliability matters. My role spans discovery, architecture, implementation, and the final production details that make software useful.
Applied AI systems
Agentic workflows, LLM evaluation, retrieval pipelines, human-in-the-loop review
Full-stack products
React, Next.js, Node.js, Express, Flask, Angular, Vue, PostgreSQL
ML & data
Python, TensorFlow, Pandas, NumPy, scikit-learn, feature engineering
Production infrastructure
AWS, Google Cloud, Docker, Redis, Celery, MongoDB, observability
Tech Lead, Google Developer Group
Leading hands-on workshops in scalable cloud architecture, system design, and AI-enabled development at the University of Toronto.
Honours BSc, Computer Science
University of Toronto · Dean's List Scholar · 3.77/4.0 · Expected June 2027