Open to opportunities

jay patel

@jaypatel11

Message

Senior AI Engineer specializing in production LLM, RAG, and agentic systems for regulated healthcare and life sciences.

United States

Message

What I'm looking for

I’m looking to build trustworthy, explainable AI in regulated, data-intensive environments—owning systems from data pipelines to low-latency inference—while partnering cross-functionally to deliver measurable quality, compliance, and long-term maintainability.

I’m a Senior AI Engineer with 10+ years designing and deploying production ML and LLM systems across healthcare, life sciences, and enterprise SaaS. I focus on scientific accuracy, domain relevance, and regulatory compliance—especially where explainability and auditability matter.

In my recent role, I build end-to-end GenAI systems with practical, measurable outcomes: agentic workflows for investigations, schema-guarded NL→SQL pipelines, and RAG architectures that improve answer relevance while reducing hallucinations. I also deliver guided reasoning for extraction and reporting from complex JSON artifacts, so teams can trust outputs and act faster.

Across my work, I emphasize transparency and human control—reasoning traces, structured intermediate outputs, and human-readable audit trails—so stakeholders can verify, override, and improve results. I also lead cross-functional delivery by mentoring engineers, running GenAI design reviews, and translating complex model behavior into actionable insights for non-technical teams.

Earlier, I owned optimization and hardening for edge deployments, reducing model size and latency while keeping accuracy degradation under 1%. I also integrated SHAP-based explanations and risk-tiering into clinician-facing dashboards, improving adoption and supporting value-based care outcomes for large patient populations.

Experience

Work history, roles, and key accomplishments

Current

Senior AI Engineer

Current

Vectra AI

Jul 2023 - Present (2 years 10 months)

Designed and implemented agentic investigation workflows for cybersecurity campaign analysis, reducing analyst triage time by an estimated 35–50%. Built schema-guarded NL→SQL, hierarchical RAG with reranking, and transparent reasoning/audit trails to improve answer relevance and reduce hallucinations.

Cross Encoder Re Ranking RAG NL→SQL Pydantic Prompt Engineering Explainable AI

Machine Learning Engineer

Qualcomm

Jun 2019 - Jun 2023 (4 years)

Owned end-to-end optimization of vision and sequence models for Snapdragon edge deployment, reducing model size by ~45% and inference latency by ~30% with <1% accuracy degradation across multiple chipset targets. Built scalable training/evaluation pipelines and production-hardened CI/CD-gated ML code, reducing integration bugs by ~60%.

INT8 Quantization Python Apache Spark A B Testing C++ Inference CI CD

Data Scientist

Innovaccer

Jun 2015 - May 2019 (3 years 11 months)

Integrated SHAP-based explanations into clinician-facing dashboards, increasing clinician adoption by ~35% within 6 months after launch. Built and validated risk, readmission, and engagement models on claims/EHR data, improving data completeness from ~65% to 90%+ and achieving AUC 0.82+ for patient risk stratification.

SHAP Explainability Risk Modeling Claims Precision@K Optimization AUC Evaluation