Cédric Kessahou

Cédric KessahouI find the real bottleneck. I build the tool. I put it in the hands of the people who use it.

Bilingual French–English developer in Toronto: I write the code, in data, AI and web, and I talk to the people it's for. Open to Solutions Engineer, Data/BI Analyst and ML/AI Engineer roles.

See the projectsDownload my CV
  • Open work permit, no sponsorship required
  • Toronto, hybrid, or remote anywhere in Canada
  • French · English · Mandarin (HSK 4)

Selected work

Each project starts with a real problem. The first one cut a FedEx team's daily data entry from ~3 h to ~20 min.

Parcel Vision extracting a parcel label: tracking number, recipient, postal code, dimensions and weight, each ticked. Sample data.Product film

Personal initiative at FedEx, Chicoutimi, approved and supported by my manager

Parcel Vision

The agent takes a photo of a parcel; the data lands in the subcontractor's spreadsheet. Built at FedEx, where three agents were retyping ~300 parcels a day.

−89%daily manual data entry, from ~3 h to ~20 min

Creator. Problem framing, build, evaluation, deployment and user training2025 – 2026

  • Python
  • Streamlit
  • Claude Vision API
  • Qwen2.5-VL + LoRA
  • EasyOCR
  • PostgreSQL
  • Docker
  • Caddy
  • VPS
Read the case study
Habita home page in French: 'Trouve ton chez-toi, jamais seul·e dans la recherche', with a card showing a 92 match score for a listing.

Startup. Began at the CEE-UQAC “Boost ton idée!” bootcamp, where the team placed second

Habita

A rental-matching startup that pairs compatible roommates and connects landlords with the right tenants. It started at a university entrepreneurship bootcamp and became a real product.

110+respondents in the market research

Founder & Product/AI LeadFall 2025 – present

  • Python
  • FastAPI
  • Pydantic
  • REST API
  • Self-hosted servers
Read the case study
WeSpeak landing page: 'Learning words is easy. Not forgetting them is hard.', with the app's daily plan on a phone screen.

Personal product. I built it because no existing app matched the way I learn

WeSpeak

A progressive web app for remembering the English words you meet every day, with spaced repetition and AI-generated examples.

Creator. Product, SQL schema, front end, AI pipeline, deploymentJune 2026 – present

  • React
  • Vite
  • TypeScript
  • Supabase (Auth + PostgreSQL)
  • Node / Express
  • GitHub Actions
  • Vercel
Read the case study
How it works
  1. QuestionA CSV file and a question in plain English, typed or spoken (transcribed by Whisper).
  2. AgentA local LLM (Gemma, served by Ollama) plans the analysis and chains up to four tool calls.
  3. MCP serverFastMCP exposes seven data tools built on Pandas. Results move between tools as dataset handles.
  4. AnswerThe model explains the result in plain language, without ever seeing the raw rows.

Gradio interface · runs fully on the local machine, no cloud API

Personal project, open source (MIT licence)

Mini Data Science Assistant

An agentic assistant for CSV files. Ask in plain English, typed or spoken; a local LLM plans the analysis, chains the right tools and explains the result.

Creator. Architecture, code and agent design2026

  • Python
  • Ollama
  • Gemma
  • FastMCP
  • Pandas
  • Matplotlib
  • Whisper
  • Gradio
Read the case study

How I work

The same three steps run through everything I've built, from a FedEx dock to a startup.

  1. Find the real bottleneck

    I start where the work happens, next to the people doing it, and look for the task that eats the hours.

    At FedEx, three agents were retyping the details of ~300 parcels a day, by hand.

  2. Build the tool

    I build the smallest thing that removes the problem, and I measure it before I trust it.

    I tested my own fine-tuned vision model against the Claude API on 69 real labels, then shipped the more accurate one.

  3. Put it in people's hands

    A tool only counts once people use it. I train them, listen, and adjust.

    I trained the agents on Parcel Vision myself. Daily data entry went from ~3 h to ~20 min.

Also built

Client websites

Jeune Trans

Animated website for a freight-forwarding and customs-clearance company in Lomé, Togo, with an interactive 3D globe.

  • HTML
  • CSS
  • JavaScript
  • Three.js
  • Vercel

Kem Digital

Website for a consulting and software studio in Abomey-Calavi, Benin, built around an interactive 3D scene and covered by end-to-end tests.

  • Astro
  • Three.js
  • GSAP
  • Playwright
  • Vercel

Small tools

  • Job-application tracker

    A CSV log turned into an Excel workbook by Python, plus a Google Apps Script version, with a KPI dashboard: response rate, interview rate, and conversion by channel and by résumé version.

    • Python
    • openpyxl
    • Google Apps Script

  • Personal productivity dashboard

    A single HTML page, saved in the browser, built around a weekly "rule of three".

    • HTML
    • JavaScript
    • localStorage

Experience

  1. 2024 – present

    Logistics Coordination & Customer Service

    FedExToronto, ON · previously Chicoutimi, QC

    • Built the shift schedules and dispatched a team of ~18 package handlers processing 2,500–3,000 parcels a day.
    • Coordinated cargo operations and managed supply and equipment inventory.
    • Owned follow-up on problem parcels with customers, subcontractors and vendors, from diagnosis to resolution.
    • Spotted a data-entry bottleneck and built Parcel Vision to remove it, with my manager's approval and support.
    • In Toronto since late August 2026.

  2. Jul 2025 – Apr 2026

    Production Operator

    Scepter AluminiumChicoutimi, QC

    • Aluminum products production line, alongside full-time studies.

  3. Fall 2024

    Teaching Assistant, Programming Fundamentals

    Université du Québec à Chicoutimi (UQAC)Chicoutimi, QC

    • Led C++ lab sessions, tutored students one-on-one and graded their work.

  4. 2020 – 2022

    Executive Assistant to the General Manager

    CAAM BTPCotonou, Benin

    • Handled payroll, subcontracting contracts and bookkeeping; ran meetings and kept management, subcontractors and partners in sync.

Volunteering

  • Centre d'Action Bénévole du Saguenay Volunteer. Received the "Nouvelle Voix" certificate of recognition.

Skills

ML / AI

  • Python
  • Pandas
  • NumPy
  • scikit-learn
  • TensorFlow / PyTorch
  • Computer vision
  • OCR
  • LLM integration

With clients

  • Solution scoping
  • Technical demos
  • Requirements gathering
  • Training & documentation
  • Escalations & exception handling
  • Bilingual FR/EN support

Data / BI

  • SQL
  • PostgreSQL
  • MySQL
  • Power BI
  • R

Web

  • TypeScript / JavaScript
  • React
  • Vite
  • REST APIs
  • Supabase

Deployment & tools

  • Docker
  • Coolify
  • VPS
  • Git / GitHub
  • GitHub Actions

Other

  • C++

Education

  • April 2026

    Bachelor of Computer Science, Data Science & Business Intelligence

    Université du Québec à Chicoutimi (UQAC), Chicoutimi, QC

  • 2019 – 2022

    Professional Bachelor's in Chinese Language, Translation & Interpreting

    Confucius Institute, Université d'Abomey-Calavi (UAC), Benin

Languages

French
Native
English
Professional working proficiency
Mandarin Chinese
HSK 4

Certifications & training

  • In progress

    Microsoft Certified: Azure AI Apps and Agents Developer Associate (exam AI-103)

    Microsoft · Preparing for the exam.

  • In progress

    MCP Course — Model Context Protocol and AI agents

    Hugging Face

  • Planned

    Lead With AI: Adapt, Implement & Transform Your Organization

    MasterClass Certificates, with Microsoft

  • Planned

    Scaling AI Strategy: Build, Launch, Drive What's Next

    MasterClass Certificates, with NVIDIA

Right now

September 2026

Building
WeSpeak's verb-tense practice, and its iOS and Android apps with Capacitor.
Building
Habita's matching engine and product, with a team of six.
Learning
The Model Context Protocol and AI agents (Hugging Face course), while preparing the Azure AI-103 exam.
Looking for
A Solutions Engineer, Data/BI Analyst or ML/AI Engineer role in Toronto or remote in Canada.

Let's talk.

Email is the fastest way to reach me. I answer in English or French.

Résumé

Full résuméSolutions engineering, data/BI and ML/AI, in one document

Profiles

  • Based in Toronto, ON. Hybrid, on-site, or remote anywhere in Canada.
  • Open work permit, no sponsorship required.