AI / Senior AI Engineer

Innovaccer
Noida
Workplace: OnsiteFull timeFunction: Data Science & Machine LearningEducation: phdSkills: ["Ability to take work from paper to prototype to production","Experimental design","Systems sense","Clear communication","Build rigorous evaluations"]

Build and own agentic language-model systems for healthcare workflows, including post-training and domain adaptation of open-weight models across model sizes. Design end-to-end model systems behind product features like document understanding, policy retrieval, reasoning, and drafting, with routing, retrieval/tool layers, verifiers, and robust production deployment. Develop training-data pipelines for HL7/FHIR/claims/PDFs, run multi-GPU training at scale, and create rigorous evaluation, safety, PHI guardrails, and latency/cost-aware inference.

Loading

Loading job details...

Preparing the role view and application actions.

FursaFursa
Innovaccer
Innovaccer
2 days ago

AI / Senior AI Engineer

✓ Verified Job

Canonical indexed version, validated from employer's careers page.

Source: Company careers pageValidated by: Fursa AI
Last checked: 5 hours agoStatus: Live

Job Summary

Build and own agentic language-model systems for healthcare workflows, including post-training and domain adaptation of open-weight models across model sizes. Design end-to-end model systems behind product features like document understanding, policy retrieval, reasoning, and drafting, with routing, retrieval/tool layers, verifiers, and robust production deployment. Develop training-data pipelines for HL7/FHIR/claims/PDFs, run multi-GPU training at scale, and create rigorous evaluation, safety, PHI guardrails, and latency/cost-aware inference.
Location: Noida
Workplace: Onsite
Employment Type: Full time
Job Function: Data Science & Machine Learning
Seniority: Mid level

Key Responsibilities

  • •Own the model portfolio behind product surfaces by selecting model sizes and composing model systems (routing/cascades, retrieval/tool layers, structured decoding, verifier models, fallbacks).
  • •Drive post-training and domain adaptation across clinical, claims, and payer-policy data using supervised fine-tuning, preference optimization, RL with verifiable rewards, continued/domain-adaptive pretraining, and distillation.
  • •Build and run large-scale distributed training jobs with multi-node training, throughput tuning, checkpoint/resume, and failure diagnosis for costly runs.
  • •Create training-data pipelines from HL7/FHIR/claims/PDFs/faxes/free text into deduplicated, decontaminated, quality-filtered, PHI-safe corpora, including mixture design and synthetic-data quality controls.
  • •Develop evaluation, safety, and guardrails (task-specific benchmarks, LLM-as-judge calibration, statistical rigor with confidence intervals, PHI detection/redaction, and audit trails) and deploy with versioning, shadow deployment, staged rollout, drift detection, and regression tests.

Key Requirements

  • •MS or PhD in Computer Science, Machine Learning, or a related quantitative field (exception for exceptional BS candidates).
  • •Depth beyond coursework: first-author publications at NeurIPS/ICML/ICLR/ACL/EMNLP or similar; meaningful open-source contributions; or research internships/models shipped and used.
  • •Fine-tuned an open-weight model and can explain parameter-efficient vs full fine-tuning decisions.
  • •Strong Python and PyTorch, with familiarity of the training/serving stack (HuggingFace, FSDP or DeepSpeed, vLLM or SGLang, or equivalents).
  • •Some exposure to multi-GPU training and sharding strategy; evidence you can take work from idea to production; senior level requires production post-training and multi-node runs (tens of GPUs) plus owning data pipelines.
Experience:HealthcareOpen sourceAgentic AI
Education:PhD / Doctorate in Computer Science, Machine Learning, or related quantitative field
Skills:Ability to take work from paper to prototype to productionExperimental designSystems senseClear communicationBuild rigorous evaluations
Tech Stack:PythonPyTorchHuggingFaceFSDPDeepSpeedVLLMSGLangCUDANCCLDistributed trainingMulti-GPU training

Company Brief

Innovaccer
Provides a healthcare data activation platform that unifies clinical, claims, and operational data to enable analytics, care management, population health, and value-based care initiatives for providers, payers, and life sciences organizations.
Industry: HealthTech
Company Size: Enterprise (1,001+ employees)
Growth: Scaleup
Valuation: Unicorn (USD 1B+)
Funding: Series D
Headquarters: San Francisco, United States
Founded: 2014
WebsiteLinkedIn