Technical Expertise

Tools: PyTorch, TensorFlow, SQL, CUDA, Weights & Biases
Domains: Bioimaging, Predictive Analytics, Computer Vision, Automation
Methodologies: Probabilistic ML, Multimodal Image Translation, CI/CD Pipelines

Industry-Aligned Experience

For Life Science & Healthcare

• GSK Pharma R&D:
  Designed diffusion model-based pipelines for virtual staining, reducing processing time from weeks to minutes. Published at ISBI/ECCV.
• Deep Learning @ Marine Biological Lab:
  Created open-source tools for 3D nuclei segmentation, adopted by Chan Zuckerberg Biohub. Trained biologists in AI adoption.

For Tech & Open Source

• The Alan Turing Institute:
  Core developer of Scivision, expanding its model library and CI/CD pipelines. Collaborated with Google DeepMind/Microsoft.
• Workshops: Led sessions on uncertainty quantification (Turing Institute) and generative models (MBL), emphasizing reproducibility.

For Finance & Business Analytics

• Haleon (GSK Consumer Healthcare):
  Built LSTM-based forecasting pipelines, improving market prediction accuracy for product launches.
• MSc Expertise: Optimization, network analytics, and operational automation (Imperial College London).

Selected Achievements

• Reduced trained model count by 40% for large-scale deployment at GSK through data-centric generalization strategies.
• Improved forecast accuracy for Haleon using E2E deep learning pipelines, enabling data-driven brand decisions.
• Launched open-source repositories (virtual staining, Scivision) with 1,200+ GitHub stars collectively.

Publications & Talks

• “Evaluation of Virtual Staining for High-Throughput Screening” (ISBI 2023)
  Demonstrated ML-driven efficiency gains for industrial microscopy.
• “Can Virtual Staining Generalize?” (ECCV 2024 Workshop)
  Identified data-centric approaches for cross-domain model robustness.
• Workshop on Uncertainty Quantification (Alan Turing Institute)
  Facilitated industry-academia dialogue with Google DeepMind/Microsoft.

Let’s Connect!

Whether you’re exploring AI in healthcare, scalable ML pipelines, or data-driven forecasting, I’d love to discuss how my expertise can drive innovation in your field.

📩 Contact: smtonks2712@gmail.com 🏃 Fun Fact: Training for the Paris Marathon 2025 while learning vinyl DJing!