ML Research Engineer Jobs

View detail

Design and ship advanced ML systems, especially LLM-driven agents and self-improving workflows. Build robust data and training pipelines, enable fast experimentation, and ensure models and agents continuously improve in production. Build LLM-based agents, tool-using workflows, and autonomous self-improvement loops. Design, train, and evaluate ML models across NLP/LLM, vision, and retrieval domains. Develop data pipelines, training code, experiment tooling, and automated deployment systems. Use PyTorch for model development and W&B (or similar) for tracking experiments and lineage. Implement monitoring for performance, drift, safety, and agent behavior. Optimize inference for latency, throughput, and cost. Work closely with engineering and product teams to turn prototypes into reliable production features. Establish ML engineering best practices and mentor teammates.

View detail

As a Senior Machine Learning Researcher at AKASA, you will lead the design, training, and evaluation of large language models to address healthcare-specific challenges, focusing on advancing clinical Natural Language Understanding. You will collaborate with cross-functional teams including PhD researchers, ML engineers, and healthcare experts to integrate Human-in-the-Loop data for model improvements and explore optimization methods. Your role includes working end-to-end on model design, data creation, training, evaluation, and iteration to ensure research advances both models and real-world healthcare tasks. You will stay updated on machine learning advancements to maintain AKASA's leadership in healthcare AI, partner with healthcare experts to align models with real-world needs, contribute to high-impact publications, and support the integration of your research into AKASA's product offerings used across healthcare systems.

View detail

The Tech Lead is responsible for architecting the system and mentoring a team of three engineers while spending significant time hands-on in the codebase using Python and PyTorch. They will own the core chat loop, optimizing context windows, memory/RAG retrieval, and inference latency to ensure a seamless real-time experience. They must drive the strategy for supervised fine-tuning (SFT), reinforcement learning with human feedback (RLHF/DPO), deciding when to prompt, fine-tune, or architect new retrieval augmented generation (RAG) pipelines. They manage the "Data Engine" overseeing sourcing, labeling, and cleaning datasets to improve model steerability and multicultural performance. Additionally, they design and train custom classifiers for high-precision moderation to detect and filter non-consensual or illegal content, moving beyond binary safe/unsafe flags to enable nuanced, context-aware moderation systems within an uncensored, NSFW environment.

View detail

Lead the LLM team of 3 engineers by acting as both architect and hands-on coder, writing production code in Python/PyTorch, and mentoring the team. Own and optimize the core chat loop, including context windows, memory/RAG retrieval, and inference latency to ensure a real-time user experience. Drive the strategy for supervised fine-tuning (SFT) and RLHF/DPO (Preference Optimization), deciding when to prompt, fine-tune, or design a new RAG pipeline. Manage the data engine responsible for sourcing, labeling, and cleaning datasets to improve model steerability and multicultural performance. Architect and build sophisticated, context-aware moderation classifiers and alignment strategies to detect and filter non-consensual or illegal content in an explicit environment, moving beyond binary safe/unsafe flags.

View detail

The Tech Lead will ship code and lead from the front by architecting the system and mentoring the team while spending significant time hands-on in the codebase using Python and PyTorch. They will own the core chat loop by optimizing context windows, memory/retrieval-augmented generation (RAG) retrieval, and inference latency to ensure a seamless, real-time experience. They will own the model lifecycle by driving strategy for supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF/DPO), deciding when to prompt, fine-tune, and architect new RAG pipelines. They will manage the sourcing, labeling, and cleaning of diverse datasets to improve model steerability and multicultural performance. Additionally, they will architect high-precision moderation by designing and training custom classifiers to detect and filter non-consensual or illegal content in an explicit environment and create nuanced, context-aware moderation systems beyond binary safe/unsafe flags.

View detail

The Tech Lead will ship production code and lead the LLM team of 3 engineers by acting as both architect and mentor. Responsibilities include owning the core chat loop by optimizing context windows, memory/RAG retrieval, and inference latency for a real-time experience. The role involves driving strategy for supervised fine-tuning (SFT), RLHF/DPO preference optimization, managing data sourcing, labeling and cleaning to improve model steerability and multicultural performance. Additionally, they will architect high-precision moderation systems by designing and training custom classifiers to detect and filter non-consensual or illegal content in an explicit environment, moving beyond binary safe/unsafe flags to nuanced, context-aware moderation systems.

View detail

Lead the LLM team, owning the architecture, training, and deployment of models powering the core product. Act as a player/coach by architecting the system, mentoring the team, and actively writing production code primarily in Python/PyTorch. Optimize the core chat loop focusing on context windows, memory/RAG retrieval, and inference latency to deliver a seamless real-time user experience. Drive the strategy for model lifecycle management including supervised fine-tuning (SFT), reinforcement learning with human feedback (RLHF), and direct preference optimization (DPO), deciding when to prompt, fine-tune, or architect new retrieval-augmented generation pipelines. Manage the data engine involving sourcing, labeling, and cleaning datasets to enhance model steerability and multicultural performance. Architect and build high-precision moderation systems by designing and training custom classifiers to detect and filter non-consensual or illegal content in an explicit environment, moving beyond binary safe/unsafe flags towards nuanced, context-aware moderation.

View detail

The Tech Lead will act as a player/coach, architecting the system and mentoring the team while spending significant time hands-on in the codebase (Python/PyTorch). They will own the core chat loop, optimizing context windows, memory/RAG retrieval, and inference latency to ensure a seamless, real-time experience. They will drive the strategy for supervised fine-tuning (SFT) and reinforcement learning with human feedback/preference optimization (RLHF/DPO), deciding when to prompt, fine-tune, or architect a new RAG pipeline. They will manage the data engine overseeing the sourcing, labeling, and cleaning of diverse datasets to improve model steerability and multicultural performance. Additionally, they will architect high-precision moderation by designing and training custom classifiers to detect and filter non-consensual or illegal content within an explicit environment and create nuanced, context-aware moderation systems beyond binary safe/unsafe flags.

View detail

The Tech Lead will take the helm of the LLM team, owning the architecture, training, and deployment of the models powering the core product. Responsibilities include writing production code, defining alignment strategies, and shipping features used by millions globally. They will act as a player/coach, architecting the system and mentoring the team while spending significant time hands-on in Python/PyTorch code. They will own the core chat loop by optimizing context windows, memory/RAG retrieval, and inference latency to ensure a seamless real-time experience. They will drive the strategy for Supervised Fine-Tuning (SFT) and Reinforcement Learning with Human Feedback/Direct Preference Optimization (RLHF/DPO), managing when to prompt, fine-tune, or architect new RAG pipelines. Additionally, they will oversee the sourcing, labeling, and cleaning of diverse datasets to improve model steerability and multicultural performance. Another key responsibility is to architect high-precision moderation by designing and training custom classifiers to detect and filter non-consensual or illegal content in an explicit environment, moving beyond binary safe/unsafe flags to nuanced, context-aware moderation systems.

View detail

As Tech Lead of the LLM team, you will architect the system and mentor your team while spending significant time hands-on coding in Python/PyTorch. You will own the core chat loop, optimizing context windows, memory/RAG retrieval, and inference latency to enable a real-time user experience. You will drive the strategy for supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF/DPO), deciding when to prompt, fine-tune, or build new RAG pipelines. You will manage the data engine by overseeing sourcing, labeling, and cleaning diverse datasets to enhance model steerability and multicultural performance. Additionally, you will design and train custom classifiers for nuanced, context-aware moderation to detect and filter non-consensual or illegal content in an explicit environment, moving beyond simple binary safety flags to build a precise moderation system.