We are building a new AI infrastructure product in stealth.

The work sits at the intersection of applied machine learning, model adaptation, benchmarking, code intelligence, and AI systems.

We are looking for a Founding Machine Learning Engineer to lead fine-tuning, dataset creation, and model evaluation from the ground up. This is a zero-to-one role with significant ownership over experimentation, evaluation methodology, and technical direction.

• Location: Remote (India)
• Employment type: Full-time
• Joining: As soon as possible
• Compensation: Based on experience and fit

• Create and curate high-quality datasets for model training and evaluation
• Run supervised fine-tuning, LoRA, and other model-adaptation experiments
• Design private, held-out evaluations based on real product use cases
• Select, configure, and adapt relevant public benchmarks
• Establish reproducible baselines and experimental controls
• Define train, validation, calibration, and test splits
• Detect and prevent dataset, repository, and task leakage
• Compare base, prompted, adapted, and fine-tuned models
• Define meaningful metrics and failure taxonomies
• Run automated and human evaluation workflows
• Analyse statistical significance, uncertainty, and variance
• Measure model quality, latency, token usage, and cost
• Investigate model failures and identify opportunities for improvement
• Communicate experimental results clearly and use them to guide technical decisions

• Strong applied machine learning and ML engineering fundamentals
• Strong Python skills
• Experience fine-tuning or adapting language models
• Experience creating, cleaning, and validating datasets
• Experience designing and running evaluations for language models
• Familiarity with supervised fine-tuning, LoRA, and related techniques
• Strong understanding of experimental design and statistics
• Ability to run rigorous and reproducible ML experiments
• Ability to reason carefully about data leakage, bias, variance, and misleading results
• Ability to work independently and take ownership of ambiguous technical problems
• 2+ years of relevant ML engineering, applied ML, or research-engineering experience, or equivalent evidence of exceptional ability

• Experience evaluating LLMs, coding models, or autonomous agents
• Experience fine-tuning open-source language models
• Experience with PyTorch, Hugging Face, vLLM, or managed training platforms
• Experience with code-generation or repository-level benchmarks
• Experience building contamination-resistant evaluations
• Experience with model-based graders or human evaluation
• Experience working with synthetic data generation and filtering
• Open-source, research, or production work involving fine-tuning, datasets, or model evaluation

Further product and architecture details will be shared during the interview process.