Senior AI Model Builder (Foundation Models & Scientific ML)
Engineering AI/Artificial Intelligence Machine Learning Python
年収
Location
Benefits
13ヶ月目の給与
その他の福利厚生
Job Overview And Responsibility
We are looking for a Senior Backend Engineer to help build and scale the backend engine of our drug-discovery Super Intelligence. You will: - Design and build the APIs, services, and data flows that power AI agents, models, and lab workflows. - Work closely with AI researchers, data scientists, and biologists to turn ideas into running systems. - Help shape how our PD and CNS foundation models are served, observed, and improved over time. - This is a hands-on role with high ownership. You will write code, make architecture decisions. KEY RESPONSIBILITIES 1. Lead PD-focused foundation model development - Own the strategy to fine-tune / distill LLMs into a PD-specialized model for scientific reasoning, evidence-grounded answers, and safe tool usage. - Design instruction and preference datasets, including curation, synthetic data generation, and quality gates. - Implement alignment techniques to reduce hallucinations and increase reliability in biomedical contexts. 2. Run serious experiments, not toy demos - Define benchmarks, offline evaluation suites, and rubrics (accuracy, calibration, robustness, hallucination rate, tool-call correctness). - Run ablations and deep error analysis; identify failure modes and drive targeted model/data fixes. - Set up continuous evaluation and regression checks to protect quality across model versions. 3. Ship models into production with engineers, scientists, and lab - Work with MLOps to package models for GPU inference, version them, and monitor them in production. - Define stable model interfaces (schemas, metadata, confidence fields, provenance hooks) so downstream services remain reliable. - Work with computational biologists to ensure model outputs make biological sense, not just statistical sense. - Incorporate feedback from lab experiments to update datasets, labels, and training strategies. - Maintain a clean experiment registry: configs, datasets, metrics, and reproducible training runs.
Required Skills and Experience
- 4+ years working on machine learning / deep learning (industry, research lab, or strong personal projects). - Excellent fundamentals in deep learning, probability/statistics, and evaluation methodology. - Strong skills in Python and at least one major DL framework. - Hands-on experience training or fine-tuning large models (LLMs, vision models, graph models, or multimodal models). - Comfortable with data pipelines: cleaning, splitting, augmenting, and loading large datasets efficiently. - Able to read ML papers and turn ideas into running code and experiments. - Curious about biology and drug discovery, even if you are still learning the domain. - Enjoy working in a fast, messy, ambitious environment where specs are not always final, and you help design the answer.
Why Candidate should apply this position
- Ownership of core intelligence: You’ll lead the modeling direction for Parkinson’s Disease-focused foundation models and scientific predictors that sit at the heart of our end-to-end discovery engine—not side experiments. - High-impact mission: Your work directly affects how quickly we can generate and prioritize better Parkinson’s Disease drug candidates, with real-world downstream validation. - GPU-first build environment: Access to modern GPU infrastructure and an engineering stack designed for training + high-throughput inference, with strong MLOps support for deployment. - End-to-end shipping culture: You’ll take models from idea → experiment → evaluation → production. We value measurable impact, reproducibility, and decision- grade reliability. - Annual leave - SHUI and Health Insurance - Bonuses and 13th-month salary. - Other benefits are considered on a case-by-case basis
Preferred skills and experiences
- Experience with scientific ML: molecules (SMILES, graphs), proteins, or biomedical texts. - Familiarity with NVIDIA AI tools (BioNeMo, NeMo, Triton, NIM) or similar. - Experience with vector search and embeddings (pgvector, FAISS, Pinecone, Weaviate). - Built AI experiment pipelines with tracking tools (Weights & Biases, MLflow, custom dashboards). - Familiarity with knowledge graphs or graph neural networks (GNNs). - Publications, open-source contributions, or a strong project portfolio in ML/AI.
Report to
Management team
Interview process
3 rounds: Phone screen -> Assignment -> Technical round
