Anis Abdeladim

VALNET INC. · MONTRÉAL

Data Scientist

June 2026 — Present

Sole data scientist for one of Valnet's lead-generation brands, owning the full path from collection to shipped product: Cloud Run collectors, BigQuery pipelines, the analysis built on them, and the internal application surfaces used daily by editorial and leadership.
Shipped five production systems in the first six weeks, including a newsletter reporting suite stitching a historical email-platform load to a live feed over a 1.4M-row analytics backfill, a governed lead-attribution ledger, and a near-real-time news-signal watcher using LLM classification with two-source confirmation.
Works in causal inference and experimental design — fixed-effects models, confound control, conclusions gated on prospective A/B rather than observational lift — alongside embeddings and density clustering calibrated against ground truth, and judge-based evaluation harnesses for LLM components.

ARTELYS CANADA

May 2025 — Jan. 2026

Optimized production forecasting by training XGBoost regression models on technical transformer data, achieving a 50% reduction in prediction error (MAE/RMSE).
Bridged the gap between AI and engineering by implementing feature importance analysis to provide model explainability for non-technical stakeholders.
Architected and deployed a Python application to parse complex engineering files and visualize predictive insights, streamlining decision-making for factory leads.

DR. KARIM JERBI'S LAB · UNIVERSITÉ DE MONTRÉAL

May 2024 — Aug. 2024

Engineered an end-to-end ML pipeline to classify MEG neuroimaging data, successfully identifying neural signatures associated with facial familiarity.
Processed high-dimensional time-series data using MNE-Python: signal filtering, artifact rejection, and supervised learning classifiers.
Synthesized research findings into a formal scientific report and presented at the Unique Student Symposium 2024, CERVO Brain Research Center.