Research-Oriented Machine Learning Engineer

We are hiring a research-oriented Machine Learning Engineer / Software Engineer to support an active AI/ML engagement focused on synthetic persona R&D. This is a Python-based project that is already well underway, and we are looking for someone who can join the existing team, contribute quickly, and collaborate closely with another ML engineer already assigned to the project. This role is ideal for someone who is comfortable operating at the intersection of applied machine learning, LLM experimentation, statistical reasoning, and production-oriented Python development. You should be able to reason through ambiguous research problems, design and evaluate experiments, and help turn promising approaches into reliable implementation. Initial commitment is expected to be approximately 20 hours per week. You will work under the team’s project manager and collaborate directly with the existing ML engineer and broader team. Responsibilities * Support ongoing R&D for a synthetic persona / synthetic respondent system * Design, implement, and evaluate ML experiments in Python * Work with LLM-based workflows involving fine-tuning, instruction tuning, prompting, and evaluation * Help improve model behavior across zero-shot, few-shot, and task-adaptive settings * Analyze model outputs using sound statistical methods and structured evaluation approaches * Assist with dataset preparation, labeling strategy, synthetic data generation, and quality assessment * Collaborate with the existing ML engineer to refine approaches, debug experiments, and document findings * Translate research ideas into maintainable Python code that can fit into the broader project architecture * Communicate progress clearly through async updates, documentation, and project management workflows Required Background * Strong Python engineering experience * Applied machine learning experience, ideally in NLP, LLMs, or generative AI systems * Experience with LLM fine-tuning, supervised fine-tuning, instruction tuning, or other post-training workflows * Familiarity with zero-shot and few-shot learning, prompt-based evaluation, and model benchmarking * Strong grounding in statistics, experimental design, sampling, distributions, regression/classification metrics, and evaluation methodology * Ability to work through ambiguous research problems without needing every requirement fully specified upfront * Experience working collaboratively with technical teams in an async or part-time contractor environment Nice to Have * Experience with synthetic data, synthetic personas, survey simulation, market research, behavioral modeling, or agent-based systems * Familiarity with Hugging Face, PyTorch, OpenAI APIs, LangChain/LangGraph, or similar AI/ML tooling * Experience evaluating LLM outputs for consistency, realism, bias, calibration, or task performance * Background in data science, computational social science, statistics, econometrics, or survey methodology * Prior experience on research-heavy startup or applied AI projects Engagement Details * Contractor role * Approximately 10–15 hours per week initially * Python-based AI/ML project * Existing project with an active team already in place * You will work with another ML engineer and report into the project manager * Remote-friendly, async-friendly workflow Ideal Candidate The ideal candidate is not just an ML engineer who can run models, but someone who can think like a researcher and build like a software engineer. You should be comfortable forming hypotheses, testing them rigorously, interpreting results, and then turning those learnings into practical improvements to the system. We are especially interested in candidates who have worked on LLM adaptation, model evaluation, synthetic data, or research-heavy AI systems where the answer is not obvious at the start. Please submit a thoughtful proposal that reflects your actual hands-on experience, and answer the screening questions specifically. Generic AI-written proposals, copy/paste responses, or answers that do not directly address this project will not be considered.