Principal Scientist — AI/ML for Life Sciences

Bencos Research Solutions Pvt. Ltd.; India; On-site; INR 4,500,000 - 7,000,000; Full time

Job summary

Bencos Research Solutions Pvt. Ltd. seeks a Principal Scientist to own research problems from framing to deployment in biology and healthcare. Lead end-to-end ML efforts, fuse omics, imaging, structural and clinical data, and build robust models for small data regimes.

You will establish rigorous evaluation, benchmarks, and validation protocols while mentoring scientists and guiding research strategy in a fast-paced, ambiguous environment.

Qualifications

• PhD + 5 yrs experience or MS + 10 yrs experience in CS, Computational Biology, Physics, or related field.
• Proven history of delivering production AI systems.
• Deep proficiency in multi-modal AI with hands-on experience in at least two: Transformers, GNNs, Diffusion/Flow-Matching, or SE(3)-equivariant models.
• Foundational ML: Self-supervised learning, representation learning, and fine-tuning for low-data regimes.
• Experience designing and implementing agentic workflows to automate complex research or decision-making processes.
• Fluent in Python and PyTorch/JAX; strong emphasis on clean, scalable code.
• Thrives in ambiguous environments with rapid prototyping and concise cross-domain communication.

Responsibilities

• End-to-End ML: Architect, train, and deploy production ML models tailored to complex life science datasets.
• Multi-Modal AI: Fuse omics, imaging, structural, and clinical data to drive biological discovery.
• Robust Modeling: Build resilient architectures designed to overcome small sample sizes, batch effects, and noisy data.
• Rigor & Validation: Establish rigorous benchmarks, evaluation metrics, and validation protocols.

Skills

Deep multimodal AI

Transformers

GNNs

Diffusion/Flow-Matching

SE(3)-equivariant models

Self-supervised learning

Representation learning

Fine-tuning for low-data

Python

PyTorch/JAX

MLOps

Cloud (AWS/GCP/Azure)

Education

PhD + 5 yrs experience

MS + 10 yrs experience

Tools

Python

PyTorch

JAX

AWS

GCP

Azure

Job description

We are a small, agile team building AI models to solve complex problems in biology and healthcare. As a Principal Scientist, you will own research problems end-to-end—from framing and prototyping to validation and deployment. You will anchor our technical efforts, transforming ambiguous biological questions into actionable ML solutions.

What you will Do

• End-to-End ML: Architect, train, and deploy production ML models tailored to complex life science datasets.
• Multi-Modal AI: Fuse omics, imaging, structural, and clinical data to drive biological discovery.
• Robust Modeling: Build resilient architectures designed to overcome small sample sizes, batch effects, and noisy data.
• Rigor & Validation: Establish rigorous benchmarks, evaluation metrics, and validation protocols.

Requirements

• Background: PhD + 5 yrs experience or MS + 10 yrs experience in CS, Computational Biology, Physics, or related field. Proven history of delivering production AI systems.
• Technical Expertise: Deep proficiency in multi-modal AI with hands-on experience in at least two: Transformers, GNNs, Diffusion/Flow-Matching, or SE(3)-equivariant models.
• Foundational ML: Mastery of self-supervised learning, representation learning, and fine-tuning for low-data regimes.
• Agentic Frameworks: Experience designing and implementing agentic workflows to automate complex research or decision‑making processes.
• Engineering: Fluent in Python and PyTorch/JAX; strong emphasis on clean, scalable code.
• Mindset: Thrives in ambiguous environments with rapid prototyping and concise cross‑domain communication.

Strongly preferred

• Experience with omics data (NGS, transcriptomics, proteomics).
• Expertise in MLOps, cloud infrastructure (AWS/GCP/Azure), and large-scale model training.
• Background in regulated or GxP environments.
• Collaboration: Partner with cross‑functional domain experts to translate scientific questions into tractable ML framing.
• Technical Leadership: Guide research strategy, mentor scientists, and uphold engineering best practices.
• Scientific Impact: Drive IP creation, high‑impact publications, and foundational strategy.