Data Scientist
- Mississauga, ON, Canada
Full-time
Applications open in a new tab.
Job details
- Job type
- Full-time
- Apply by
- Posting language
- English
- Posted
Original job description
About the Client:
Our client is a Data Security Platform startup that protects sensitive data before it ever reaches an app, a pipeline, or an AI model.
Their product is used by large enterprises across Finance, Insurance, Telco, and Healthcare to reduce risk, simplify compliance, and make exposed data unusable.
They’re bootstrapped and have crossed the $15M ARR mark this year. The mission now is to enable data security in a world where Agentic AI tools are embedded within enterprise systems and have more access to private data than ever before.
The Opportunity:
- You will be the company’s first Data Scientist, reporting directly to the CTO and becoming a founding member of the AI team.
- You will build the AI-driven discovery and classification at the core of the product: ML and LLM-based classifiers, embedding similarity, behavioural models, and context-aware sensitivity detection.
- You will post train and fine tune open source LLMs to power agents in narrow security domains, taking models to production class precision.
- You will own models through the full lifecycle, from synthetic datasets to ensembles at scale, drift, and retraining.
The Skills/Requirement Needed to Succeed:
- You have built and shipped models to production across classical ML, deep learning, and LLMs.
- You have hands on experience post training and fine tuning LLMs with RL, SFT, and verifiable rewards.
- You have built behavioural, classification, or anomaly detection models. Fraud, threat detection, or data security (DSPM, DLP, CASB) experience is a strong plus.
- You have built production pipelines and served LLMs at scale.
- You are able to commute to an office in Mississauga 1 to 2 days a week. Fully remote will be considered for the right candidate.
Martyn Bassett Associates:
Martyn Bassett Associates is a team of recruiters solving the ambitious goals of startups and scale-ups by recruiting world-class sales, marketing, and product management talent who deliver growth and scale. We’ve helped over 1500 candidates land their next big opportunity at companies like Shopify, Extreme Networks, Wattpad, Loopio, Readdle, ClearCo, Workleap, Invicti, Resolve Pay, The Herjavec Group, GE Energy, GE Health, Enable, RBC Ventures, Unibuddy, Cognota, Thalmic Labs (now Google Glass), Eloqua (now Oracle Marketing Cloud) to name a few.
More about this job
Responsibilities
Build AI-driven data discovery and classification capabilities, including machine-learning and LLM classifiers, embedding similarity, behavioral models, and context-aware sensitivity detection. Own models throughout their lifecycle, from synthetic data and post-training through production deployment, scaling, drift monitoring, and retraining.
Requirements
Candidates should have shipped production models using classical machine learning, deep learning, and LLMs, and have hands-on experience post-training and fine-tuning LLMs with reinforcement learning, supervised fine-tuning, and verifiable rewards. Experience building behavioral, classification, or anomaly-detection models and production pipelines for serving LLMs at scale is required; fraud, threat detection, or data-security experience is a strong plus.
Skills
- Production Machine Learning
- Deep Learning
- Large Language Models
- LLM Post-Training
- LLM Fine-Tuning
- Reinforcement Learning
- Supervised Fine-Tuning
- Verifiable Rewards
- Behavioral Modeling
- Classification
- Anomaly Detection
- Embedding Similarity
- Synthetic Data
- Production Pipelines
- Model Drift Monitoring
- Model Retraining
Visa sponsorship
Not detected in the job text
Categories
- Data & Analytics
- Technology
- Software
- Science & Research
- Security & Safety
Keywords
- Data Science
- Data Security
- Data Discovery
- Data Classification
- Machine Learning
- Deep Learning
- Large Language Models
- LLM Classifiers
- Embedding Similarity
- Behavioral Models
- Sensitivity Detection
- LLM Post-Training
- LLM Fine-Tuning
- Reinforcement Learning
- Supervised Fine-Tuning
- Verifiable Rewards
- Synthetic Datasets
- Ensemble Models
- Production Pipelines
- LLM Serving
- Model Drift
- Model Retraining
- Anomaly Detection
- Fraud Detection
- Threat Detection
- DSPM
- DLP
- CASB
- Agentic AI