Debaditya Shome

AI Researcher

Toronto, Ontario, Canada4 yrs 10 mos experience

Most Likely To SwitchAI Enabled

Key Highlights

Authored 10+ research publications with 290+ citations.
Developed innovative AI frameworks for real-world applications.
Led large-scale AI projects impacting hiring processes.

Stackforce AI infers this person is a Senior Applied AI Scientist with expertise in AI-driven solutions across various industries.

Contact

Skills

Core Skills

Artificial Intelligence (ai)Machine Learning

Other Skills

AlgorithmsAmazon Web Services (AWS)AnalyticsApache SparkApplied Machine LearningArduino IDEArtificial Neural NetworksBig DataC (Programming Language)C++Cascading Style Sheets (CSS)Cloud AdministrationCloud ComputingComputer VisionConvolutional Neural Networks (CNN)

About

I’m a Senior Applied AI Scientist at Homebase, based in Toronto, Canada. I recently completed my Master of Applied Science in Artificial Intelligence at Queen’s University, supported by a fully funded scholarship (Vector Scholar & MITACS GGF). My research interests focus on Generative Models, Self-Supervised Learning, Multi-Modality, and learning universal representations of time-series data. I’ve authored 10+ research publications, accumulating over 290 citations, with work presented at top-tier venues like AAAI and ICASSP. Beyond academic research, I bring experience in data engineering and deploying machine learning models at scale, using cloud-native and serverless infrastructure to solve real-world problems with AI.

Experience

4 yrs 10 mos

Total Experience

11 mos

Average Tenure

1 yr 11 mos

Current Experience

Homebase

2 roles

Senior Applied AI Scientist

Promoted

Sep 2025 – Present · 9 mos · Toronto, Ontario, Canada

Sole Applied AI Scientist who ideated, built, and shipped the Hiring Assistant - the first large-scale multi-agent system here at Homebase doing thousands of autonomous interviews at scale.
Setting standards in the company around evaluations of Agentic systems including use of synthetic data, ensuring coverage/diversity, and building out an internal guardrails suite, and frameworks to develop and tune LLM-as-judges per use-case.
Building a novel framework that lets frontier LLMs excel at proprietary tool-use both in terms of quality and latency.
Tons of collaboration with product, design, and eng team leading to several Agentic features!
and 100 more things : )

Large Language Models (LLM)Data ScienceDeep LearningSoftware DevelopmentResearch SkillsArtificial Intelligence (AI)+1

Applied AI Scientist 2

Jul 2024 – Sep 2025 · 1 yr 2 mos · Toronto, Ontario, Canada

Vector institute

Machine Learning Associate

Sep 2023 – Dec 2023 · 3 mos · Ontario, Canada · Hybrid

Developed a course recommendation engine featuring a retrieval and re‑ranker pipeline; fine‑tuned a bi‑encoder with MPNET‑
base‑v2 for initial retrieval and a Contriever model for precise ranking, achieving 0.96 Top‑1 F1 score on pseudo test sets.
Constructed a large‑scale annotated dataset tailored for course recommendations using prompt engineering with GPT‑4 Turbo
and LLaMA‑70B, enhancing model training and relevance in course recommendations.

Ingenuity labs research institute

Graduate Research Assistant

Sep 2022 – Jul 2024 · 1 yr 10 mos · Kingston, Ontario, Canada

Proposed and built the first diffusion model with a disentangled region‑specific denoising process, outperforming DDPMs by 86% RMSE for the task of cross modal signal-to-signal translation.
Developed a novel cross‑modal knowledge distillation framework to make unimodal speech encoders (e.g. HuBERT) explicitly
understand the linguistic & prosodic aspect of emotions. Experiments show SOTA performance on IEMOCAP of 79% W.A.
Developed a novel prompt‑tuning framework that enables pre‑trained models to incorporate sleep information alongside wear‑
able data like ECG, PPG, and EDA, with minimal paired data, achieving an 11% boost in F1 score for mood recognition.

Queen's university - faculty of engineering and applied science

Graduate Teaching Assistant

Sep 2022 – Dec 2022 · 3 mos · Kingston, Ontario, Canada

SOFT 437 - Software Performance Analysis

Odanta

Senior AI Engineer

Feb 2022 – Aug 2022 · 6 mos · India

Led the development of a hyper-personalized Movie/OTT content Recommendation System specific to Indian audience with AWS S3, Lambda, DynamoDB, S‑BERT, and PyTorch.
Achieved a 78% cost reduction by transitioning the EC2‑based ML pipeline to serverless Lambda functions.
Deployed a continuous model re‑training pipeline using implicit user feedback to dynamically adjust recommendations.

Shl

AI Intern (Research)

Oct 2021 – Jul 2022 · 9 mos · Gurugram, Haryana, India

Thompson rivers university

MITACS Globalink Research Intern

May 2021 – Aug 2021 · 3 mos

Mitacs Globalink Research Internship is a competitive initiative for international undergraduates from 14 countries. Each year, top-ranked applicants participate in a research internship in Canadian Universities.
In this research project, I worked under Dr. Omer Waqar in the intersection of Federated leaning and wireless communications.