Experience: 2–7 years · Type: Full-time, On-site (Bengaluru)

The Mission

Mandrake Bio is a protein-design company building the next generation of programmable gene-editing enzymes. Its platform designs editors from scratch, pairing generative AI with first-principles biophysics and wet-lab validation to engineer the compactness, precision, and dynamic control that the highest-impact applications in agriculture and medicine demand. Founded in 2025 and headquartered in Bengaluru, India. Learn more at mandrake.bio.

We are a VC-backed, small, high-velocity applied research lab. We don't care about credentials; we care about proof of work, first-principles thinking, and the ability to ship.

The Archetype: Who We're Looking For

You understand the math behind the magic. You didn't just join when LLMs got popular — you have deep intuition for loss landscapes, attention, representation learning, and optimization that only comes from years wrestling with raw data and architectures. Maybe you started with a biology/biotech background and went deep into ML; maybe you're a pure ML engineer who wants to bring your toolkit "home" to biology. Either way, you want to build models that design real molecular tools, not ship another CRUD feature.

The Work

You'll have deep ownership over the AI stack. You won't be fine-tuning APIs — you'll be:

• Building and training foundation models for proteins, enzymes, and nucleic acids.
• Designing generative architectures that create functional biological tools from first principles.
• Owning the end-to-end pipeline: massive-scale biological data curation → training → RLHF/DPO-style loops that align models with wet-lab reality.
• Working directly with biologists to turn code into living, breathing experiments.
• Applying your work across both therapeutic and agricultural programs — our platform is horizontal.

You're a Great Fit If You

• Have a strong track record building and evaluating ML models (pre-LLM-era depth is a plus), with real intuition for architecture and optimization.
• Are fluent in PyTorch or JAX and comfortable at scale.
• Can conceive, implement, and rigorously evaluate novel techniques — ideally at the intersection with biology.
• Think from first principles and ship fast.

Nice to Have

• Familiarity with the foundational biology of proteins and nucleic acids.