Role & responsibilities

Functional Precision Oncology • AI-Driven Drug Response

Lead Computational Biologist / ML Scientist

Foundation Models & Fine-Tuning Precision Oncology

Location: Bengaluru, India (on-site / hybrid)

Type: Full-time • Senior / Lead individual contributor with team leadership

Reports to: Founder & CEO

About Katamaran:

Katamaran Industries is building the next generation of precision oncology: an ex-vivo functional drug-response platform (patient-derived tumour tissue slices) coupled with our multimodal AI foundation-model engine that predicts drug response by integrating transcriptomics, genomics, histopathology, and chemical structure. We turn real functional tumour biology into predictive models that help match the right drug to the right patient.

The Role

We are hiring a PhD-level lead to own our proprietary drug-prediction modelling stack end-to-end with deep, hands-on mastery of foundation-model fine-tuning. You will set the technical direction, own the multimodal architecture and the fine-tuning of large biological/chemical foundation models, and build and mentor the computational biology team. This is the senior scientific owner of our core AI capability.

What You'll Own

- Fine-tuning of large foundation models (parameter-efficient methods LoRA / QLoRA / PEFT) across biological, pathology, and chemical modalities.
- The multimodal architecture: cross-attention fusion of omics, histopathology (H&E;), and drug representations into a drug-response prediction model, including robust handling of missing modalities.
- Responder / non-responder classification methodology and target transformation (e.g. pIC50), validation strategy, and correlation with clinical and functional data.
- The full training and evaluation pipeline reproducibility, experiment tracking,



and model releases on company infrastructure.
- Technical leadership: mentor and grow the comp-bio/ML team, set standards for documentation, code, and data governance, and be the whole-system integrator.
- Close collaboration with the wet-lab and clinical teams to ground models in real functional-oncology data.

Requirements (Science & ML)

- PhD in Computational Biology, Bioinformatics, Machine Learning, Computer Science, or a closely related field. An exceptional M.Tech / Master's candidate with strong, demonstrable hands-on foundation-model experience will also be considered.
- Demonstrated, hands-on expertise fine-tuning large foundation / transformer models parameter-efficient fine-tuning (LoRA / QLoRA / PEFT) in particular.
- Experience with multimodal and/or biological data — transcriptomics/genomics, medical/histopathology imaging, and/or molecular representations (SMILES / molecular graphs).
- Sound ML fundamentals: robust validation (e.g. patient-level / leave-one-dataset-out splitting), handling noisy biological data, and honest evaluation.

Engineering

- Strong Python engineering — modular, maintainable, production-quality packages, not notebook-level scripts; able to turn research prototypes into clean, testable code.
- Strong software-engineering practice — Git branching, pull requests and code review, unit/integration testing, type hints, linting, dependency management,



and CI/CD.
- Deep-learning engineering in PyTorch, with GPU/compute optimization — profiling and troubleshooting compute, memory, and performance bottlenecks.
- Comfortable in Linux/CLI environments — shell scripting, remote development, and debugging complex systems.
- Cloud and GPU development — remote GPU machines and cloud infrastructure for training, experimentation, and large-scale data processing.
- Reproducible environments and pipelines — virtualenvs / Docker, configuration-driven training, and robust data-loading training evaluation pipelines.
- Experiment tracking and model versioning — experiment tracking, dataset/version management, checkpoints, and reproducible evaluation.

Preferred / Bonus

- Direct experience with biological or chemical foundation models (e.g. scGPT, MoLFormer, UNI/CONCH, or similar) and vision transformers for pathology.
- Background in oncology, drug-response prediction, pharmacogenomics, or precision medicine.
- Peer-reviewed publications in ML-for-biology / computational drug discovery.
- MLOps / experiment-tracking and GPU/cloud training at scale, and mentoring experience.

What We Offer

- Ownership of the core AI capability of a rapid-moving precision-oncology company, working directly with the founder.
- Competitive compensation with meaningful equity / ESOP.
- Real functional tumour data and a genuinely novel scientific problem — not incremental work.
- A path to build and lead the computational team.

To apply: Send your CV, a short note on a foundation-model fine-tuning project you led (your role, choices, and results), and any relevant publications/code to [email protected] and [email protected].

📌 Computational Biologist / ML Scientist (Bengaluru)
🏢 Katamaran Industries
📍 Bengaluru