AI / Senior AI Engineer
Innovaccer6 hours ago
Noida, IndiaSenior
Responsibilities
- Build model portfolios for healthcare product surfaces, including routing, retrieval, tool layers, structured decoding, verifier models, fallbacks, and production rollout.
- Fine-tune, post-train, distill, and domain-adapt language models across a wide range of model sizes.
- Run and optimize distributed training on hundreds of GPUs using sharding, parallelism, checkpointing, mixed precision, and throughput tuning.
- Create pipelines that transform HL7, FHIR, claims, PDFs, scanned faxes, notes, and payer-policy documents into safe, filtered, decontaminated training data.
- Develop multi-step agents with tool use, planning, memory, recovery, and reliable long-horizon behavior.
- Design task-specific evaluations, calibrate LLM judges against clinical experts, and implement PHI detection, redaction, guardrails, and audit trails.
- Build efficient inference and deployment pipelines using quantization, speculative decoding, continuous batching, KV-cache strategies, and GPU utilization optimization.
- Take research ideas from paper to prototype to production and communicate results to clinicians, operators, and non-AI stakeholders.
Requirements
- MS or PhD in Computer Science, Machine Learning, or a related quantitative field; exceptional BS candidates with substantial research or open-source work may be considered.
- Demonstrated depth beyond coursework through publications, meaningful open-source ML contributions, AI-lab research experience, or models trained and shipped for real users.
- Hands-on experience fine-tuning open-weight models and choosing between parameter-efficient and full fine-tuning.
- Fluency in Python and PyTorch, with familiarity with HuggingFace, FSDP or DeepSpeed, and vLLM or SGLang or equivalent tools.
- Some multi-GPU training experience and understanding of sharding strategies.
- For senior level: production post-training experience, product model-layer ownership, multi-node training on tens of GPUs and large models, and ownership of real training-data pipelines.
- Experience with reinforcement learning for language models, evaluation methodology, data-centric research, systems work, or regulated/high-stakes domains is valued.
- Strong experimental design, research judgment, delivery ability, and communication skills are expected.
Benefits
- Access to governed longitudinal clinical and claims data and clinical and coding experts for training and evaluation.
- Budgeted compute for multi-hundred-GPU post-training runs.
- Models are deployed into four agent families already operating in enterprise health systems and payers.
- Opportunity to help define Innovaccer's in-house modeling function and shape the team as it grows.
Categories
AI ResearchML Engineering
About Innovaccer
Innovaccer is the AI infrastructure for autonomous healthcare operations, delivering better clinical and financial outcomes across health systems, payers, governments, and life sciences. Powered by the Healthcare Intelligence Platform, Innovaccer unifies enterprise data and applies AI to automate administrative work, strengthen operational performance, and drive measurable margin expansion. Organizations such as Orlando Health, Adventist HealthCare, and Banner Health trust Innovaccer to integrate intelligence into their existing infrastructure and elevate the quality of care. For more information, visit www.innovaccer.com.