Jobgether
Accountabilities: Reproduce and evaluate machine learning research methods using open-weight and API-accessible models. Design evaluation datasets, probes, scoring approaches, baselines, calibration tests, and experiment harnesses. Work directly with model weights, logits, hidden states, activations, model APIs, and inference infrastructure when required. Build and extend evaluation infrastructure covering experiment runners, judges, persistence, orchestration, reporting, and reproducibility. Turn research workflows into intuitive product experiences, including experiment configuration, execution, traces, comparisons, reports, and review workflows. Investigate how verification methods behave when models are modified through fine-tuning, merging, quantization, distillation, safety removal, or deliberate evasion. Design controlled experiments that distinguish meaningful signals from artifacts, confounders, and misleading correlations. Produce clear technical reports that separate measured evidence from interpretation and hypotheses. Deliver production-quality systems with APIs, asynchronous jobs, databases, observability, testing, deployment, and documentation. Contribute across research, experimentation, engineering, and product as priorities evolve. During the first six months, reproduce and document at least one published model-provenance or verification method, including its capabilities, assumptions, and limitations. Build a repeatable model-verification runner with versioned inputs, artifacts, metrics, and reports, and make at least one verification workflow accessible through the product interface. Run controlled experiments across base, fine-tuned, merged, quantized, and known distilled models, improving understanding of when verification methods succeed, fail, and why. Requirements: Strong Python engineering skills, with hands-on experience using PyTorch and Hugging Face Transformers. Solid understanding of machine learning evaluation, including dataset design, baselines, metrics, calibration, false positives and negatives, statistical uncertainty, and reproducibility. Ability to read ML research papers critically and implement methods from first principles rather than relying entirely on existing packages. Professional software engineering experience beyond notebooks, including APIs, asynchronous jobs, databases, logging, testing, deployment, and documentation. Familiarity with open-weight models and a practical understanding of how modern LLM inference systems operate. Ability to work across backend and frontend boundaries, with sufficient React/TypeScript knowledge to help make complex experiments and results understandable to users. Strong experimental and analytical judgment, particularly around distinguishing what evidence demonstrates from what it merely suggests. High ownership and initiative, with the ability to identify problems, propose solutions, and drive projects forward independently. Comfort working in a fast-moving startup environment where priorities can change quickly and engineers may operate across multiple functions. Experience with model provenance, fingerprinting, watermarking, distillation detection, red-teaming, safety evaluation, interpretability, or related areas is a plus. Experience with activation and representation analysis, probing, model hooks, logits, hidden states, or other model-internals techniques is advantageous. Familiarity with evaluation and inference infrastructure such as DSPy, LiteLLM, Temporal, Ray, vLLM, PostgreSQL/pgvector, or comparable technologies is beneficial. Experience with Next.js, React, TypeScript, data visualization, or experiment dashboards is a plus. Experience running and serving open-weight models on GPUs, including reasoning about latency, throughput, memory, precision, and cost trade-offs, is valuable. Experience designing adversarial evaluations or testing systems against deliberate attempts to evade detection is an advantage. A strong commitment to producing production-quality code, tests, tooling, and documentation that other engineers can confidently operate and extend. Benefits: Opportunity to work on applied machine learning at the intersection of research, experimentation, engineering, and product. End-to-end ownership across model evaluation, model internals, infrastructure, backend systems, and user-facing experiences. Exposure to modern open-weight models, LLM inference systems, and emerging ML verification techniques. A role with significant technical autonomy and the opportunity to shape both experiments and production systems. Fast-moving startup environment with evolving priorities and cross-functional collaboration. Opportunity to translate cutting-edge research into practical, measurable, and user-accessible products. The opportunity to build systems and evaluation methodologies designed to produce evidence that users can understand and trust. Location: Romania. Additional compensation, flexibility, healthcare, and other benefits may be provided according to the partner company's employment package and local arrangements. How Jobgether works: We use an AI-powered matching process to ensure your application is reviewed quickly, objectively, and fairly against the role's core requirements. Our system identifies the top-fitting candidates, and this shortlist is then shared directly with the hiring company. The final decision and next steps (interviews, assessments) are managed by their internal team. We appreciate your interest and wish you the best! Why Apply Through Jobgether? Data Privacy Notice: By submitting your application, you acknowledge that Jobgether will process your personal data to evaluate your candidacy and share relevant information with the hiring employer. This processing is based on legitimate interest and pre-contractual measures under applicable data protection laws (including GDPR). You may exercise your rights (access, rectification, erasure, objection) at any time. #LI-CL1