Skip to main content
TALENT247 // SPECIALIST AI EXECUTIVE SEARCH & TECHNICAL RECRUITMENT
ML · Machine Learning

Senior NLP & Scientific Extraction ML Engineer

Montreal, QC, Canada · Full-time · Hybrid

AI Specialism

Machine Learning

Workplace

Hybrid

Location

Montreal, QC, Canada

Compensation

CA$155k – CA$210k

annual package

Role Dossier & Specification

REF: SENIOR-NLP-SCIENTI

THE CLIENT & MANDATE

Our client is a well-funded applied artificial intelligence company rooted in Montreal's premier deep-learning research ecosystem, with an expanding European engineering presence. They build specialized, zero-hallucination tabular extraction and scientific literature synthesis models trusted by global biopharmaceutical clinical, regulatory, and market-access teams, where a single misread clinical endpoint or omitted cohort statistic is unacceptable.

They have retained Talent247 to identify a Senior NLP & Scientific Extraction ML Engineer for their core modeling group in Montreal. Sitting alongside research scientists and production platform engineers, this hire owns the domain-tuned small language models (SLMs), multimodal document parsers, and neuro-symbolic citation verifiers that turn thousands of complex PDF clinical trial studies, regulatory filings, and nested statistical tables into provenance-linked structured datasets.

KEY RESPONSIBILITIES

  • Architect, post-train, and fine-tune specialized language and vision-language models (1B–32B parameters) using SFT, DPO, and verifiable-reward RL to extract structured clinical and scientific entities from dense PDFs and complex multi-page tables.
  • Design and maintain deterministic constrained-decoding pipelines and neuro-symbolic verification layers that bind every extracted figure, confidence interval, and claim back to exact character spans and bounding-box coordinates in source documents.
  • Build high-recall, multi-stage biomedical retrieval and cross-encoder reranking pipelines capable of screening millions of peer-reviewed publications and regulatory submissions without false negatives.
  • Create rigorous, contamination-free evaluation harnesses partnered with domain biostatisticians and medical reviewers to measure cell-level extraction precision, recall, and hallucination rates prior to production promotion.
  • Collaborate closely with inference and backend engineers to distil, quantize, and serve custom models within strict enterprise latency and GPU cost budgets.
  • Diagnose systematic failure modes across OCR, layout segmentation, and long-context reasoning, turning real-world customer edge cases into high-signal synthetic and human-verified training curricula.

TECHNICAL ENVIRONMENT & OFFERING

Their stack centers on Python, PyTorch, custom VLM/layout transformers, vLLM/TensorRT-LLM serving, Ray Data, and Kubernetes-backed H100 clusters. Operating on a collaborative hybrid schedule out of downtown Montreal, our client offers CA$155,000–CA$210,000 base salary plus meaningful early-stage equity, comprehensive Canadian health and dental coverage, conference publication support, and relocation/visa sponsorship for exceptional talent.

Core Requirements

ESSENTIAL QUALIFICATIONS

  • 5+ years of hands-on machine learning and NLP engineering experience shipping custom-trained transformer models into production environments.
  • Deep fluency in Python and PyTorch, including custom training loops, mixed-precision distributed training (FSDP/DeepSpeed), LoRA/PEFT, and preference or reinforcement post-training.
  • Proven track record in structured information extraction, document layout understanding, table-parsing vision-language models, or grammar-constrained generation.
  • Strong command of retrieval-augmented architectures, dense/sparse biomedical embeddings, neural reranking, and high-recall evaluation methodology (precision/recall calibration, abstention modeling).
  • Disciplined software engineering habits across version-controlled ML pipelines, automated regression testing, and reproducible dataset lineage.

PREFERRED / BONUS EXPERIENCE

  • Prior exposure to biomedical, clinical-trial, pharmacovigilance, or scientific literature corpora (PubMed, ClinicalTrials, FDA/EMA submissions).
  • Hands-on experience with inference optimization, model distillation, or FP8/INT4 quantization using vLLM, SGLang, or TensorRT-LLM.
  • MSc or PhD in Machine Learning, Computational Linguistics, Bioinformatics, or Computer Science, or publications at ACL, EMNLP, NAACL, NeurIPS, or ICML.

Position Telemetry

Location Montreal, QC, Canada (Hybrid)
Department Machine Learning
Employment type Full-time
Workplace type Hybrid
Salary range
CA$155k – CA$210k annual package