Montreal, Canada

Ahmed Dawod

Senior ML Engineer, Data Scientist & AI Consultant

I apply machine learning to hard, high-impact problems — from large-scale e-commerce recommender systems and drug discovery to cancer diagnostics, content intelligence, and financial inclusion. 12+ years across traditional ML, statistical analysis, LLMs, Agentic AI, NLP, Computer Vision, BCI, and FinTech.

Get in touch View consulting services
LinkedIn ↗ GitHub ↗ ResearchGate ↗

17+

Years of Software Engineering

12+

Years of Machine Learning

Fulbright

Foreign Student Scholar

GLG

Expert Network Member

About

I'm a Senior ML Engineer and Data Scientist with a passion for applying the right ML solution to the right problem. Currently at Shopify, I own and lead large-scale recommender system projects for the Shop app. Previously at BenchSci I built ML and LLM systems for drug discovery in the pharmaceutical domain. I also built one of the very first NLG content writers at MarketMuse, contributed to published cancer-diagnostic research at AOA Dx, and built alternative credit-scoring models for the unbanked at PayMob.

Beyond full-time engineering, I consult independently — helping teams find the most effective ML solution for their problem, whether that's a rigorous statistical model, a classical ensemble, a recommender system, or an LLM pipeline. I don't default to the trendiest tool; I find what works. I'm also a member of the GLG Research Expert Network, available for expert calls on ML, data science, and AI system design.

I hold an M.S. in Computer Science (ML focus) from Purdue University, funded by a Fulbright scholarship. My career started early — at 20 I was already building software professionally, and at 23 my computer vision and speech recognition system for quadriplegic patients won the Microsoft Imagine Cup in Egypt.

Consulting Services

All services →
📊

Traditional ML & Data Science

The right model for your problem — gradient boosting, statistical analysis, survival models, clustering — not just the trendiest approach.

🛍️

Recommender Systems

Large-scale recommendation engines, ranking systems, and personalization pipelines for e-commerce and consumer apps.

🤖

LLM & Agentic AI

RAG pipelines, fine-tuning, evaluation frameworks, and production-grade agentic systems.

Featured Projects

All projects →

roomsplat — 3D Room Tours from Phone Video

3D Vision / Gaussian Splatting

Turns a phone walkthrough video into a navigable 3D room tour using Gaussian Splatting — fully local, no cloud or subscription. Record a room, run one command, explore the scene in any browser.

  • › End-to-end local 3D Gaussian Splatting pipeline (35–100 min on a mid-range GPU)
  • › COLMAP for camera poses, with MASt3R fallback for texture-poor walls
  • › WebGL2 browser viewer with WASD navigation and spherical-harmonics rendering
  • › Compact ~10–60 MB .splat exports
View →

videosplat — 4D Free-Viewpoint Video

4D Vision / Gaussian Splatting

Free-viewpoint replay of dynamic scenes from multi-camera video via 4D Gaussian Splatting. Auto-calibrates cameras, trains a 4D model, and generates an interactive web viewer to scrub through time and fly through the scene.

  • › Four 4D reconstruction backends: 4DGaussians, SpacetimeGaussians, Gaussian-Flow, 4D-Rotor-Gaussians
  • › COLMAP/MASt3R calibration supporting casual, unsynced, mixed-resolution capture
  • › Person masking via confidence down-weighting for moving subjects
  • › VRAM-aware design tuned for 16GB GPUs
View →

AI Assistant for Drug Discovery

Pharma / LLMs

BenchSci

An ML/LLM-powered assistant for drug discovery scientists at BenchSci, accelerating research by surfacing relevant evidence from scientific literature and proprietary data.

  • › LLM orchestration over complex, jargon-dense biomedical corpora
  • › Custom retrieval and ranking pipeline for regulated pharmaceutical data
  • › Production system used by scientists across major pharma companies

MM NLG Suite

NLP / NLG

MarketMuse

One of the very first efficient neural content-writing systems in the market, built end-to-end at MarketMuse. Generated high-quality article drafts at scale for SEO content teams.

  • › Built the training dataset by crawling and cleaning the entire web
  • › Trained a custom Transformer-based NLG model
  • › Designed domain-specific evaluation metrics for content quality
  • › Delivered full deployment pipeline and user interface

User Intent Classifier

NLP / SEO

MarketMuse

A model to classify the search intent behind URLs and Google queries, enabling smarter content strategy recommendations for MarketMuse customers.

  • › Custom labeling platform and annotation workflow
  • › Multi-class classification across navigational, informational, transactional, and commercial intents
  • › Deployed as a backend microservice serving real-time predictions

Twasol — Assistive CV & Speech System

Computer Vision / HCI

A computer vision and speech recognition system giving quadriplegic patients full control of their computers, smart home, and telephone via head movements and voice commands.

  • › Real-time head-movement tracking via computer vision for cursor and device control
  • › Speech recognition integration for phone calls and system navigation
  • › 1st Place, Microsoft Imagine Cup Egypt 2012; Top 20 worldwide

Let's work together

Whether you need an ML strategy review, a production LLM system, or an expert consultation via GLG — I'm available.

Get in touch