Jagdish Singh Bhati

AI and Data Engineering executive with 13+ years of experience leading enterprise-scale AI transformation across financial services, healthcare, and SaaS ecosystems. Proven track record of defining AI strategy, scaling global engineering teams, modernizing cloud-native platforms, and operationalizing machine learning and generative AI systems within regulated environments. Expert in building high-performance engineering organizations delivering secure, compliant, and production-grade AI platforms.

Adept at aligning technology roadmaps with business strategy, optimizing multi-million-dollar cloud portfolios, and driving measurable impact through automation, decision intelligence, and AI-led operational efficiency. Recognized for combining deep technical authority in ML/LLMs with executive-level leadership across architecture, governance, financial stewardship, and cross-functional execution.

Executive Engineering Leader – AI - Deloitte

(2025-11)

• Lead enterprise agentic AI engineering initiatives, architecting autonomous multi-agent systems for financial reporting, variance analysis, and exception management across global finance transformation programs.
• Designed and deployed an agentic orchestration framework leveraging fine-tuned LLMs (GPT-4, Llama-3) to automate financial investigations and generate AI-driven analytical reports with human-in-the-loop validation.
• Architected multi-agent collaboration systems where specialized agents (data retrieval, variance detection, root-cause analysis, report generation) communicate via structured protocols, automating 70% of investigation workflows and reducing analysis time from days to hours.
• Implemented ReAct (Reasoning + Acting) and Chain-of-Thought prompting strategies enabling agents to dynamically plan investigation workflows and synthesize findings into executive-ready variance explanations.
• Built agent evaluation frameworks measuring task accuracy, reasoning coherence, and hallucination rates across financial scenarios; conducted systematic experiments to improve agent reliability and performance.
• Established enterprise AI governance standards for autonomous agents including safety guardrails, output validation, and explainability mechanisms to ensure regulatory compliance.
• Mentored cross-functional teams on agentic AI design patterns, tool-use capabilities, and orchestration architectures for scalable autonomous systems.

Technical Leader – AI Platforms & ML Engineering - R Systems International

(2022-03 - 2025-10)

• Directed RAG (Retrieval-Augmented Generation) architecture strategy and Small Language Model (SLM) deployments for enterprise knowledge management and domain-specific AI platforms.
• Architected hybrid retrieval pipelines integrating dense retrieval (Sentence Transformers, bi-encoders) and sparse retrieval (BM25, TF-IDF) with custom reranking models, improving answer relevance by 45%.
• Led domain-specific LLM fine-tuning initiatives using parameter-efficient methods such as LoRA and QLoRA to adapt foundation models for enterprise use cases.
• Built large-scale fine-tuning pipelines using PyTorch and Hugging Face Transformers, implementing instruction tuning, supervised fine-tuning (SFT), and Direct Preference Optimization (DPO).
• Developed synthetic data generation frameworks using GPT-4 and Claude to create high-quality training datasets (50K+ instruction-response pairs), improving model accuracy by 40%.
• Designed evaluation frameworks for RAG systems measuring precision@k, answer faithfulness, contextual relevance, and generation quality using LLM-as-judge and human evaluation.
• Established MLOps / LLMOps pipelines enabling continuous experimentation, A/B testing, embedding drift detection, and production monitoring of retrieval quality.

Data Analytics Specialist - American Express

(2016-01 - 2022-01)

• Contributed to enterprise NLP and deep learning initiatives supporting real-time credit risk modeling, fraud detection, and customer intelligence platforms processing millions of transactions daily.
• Developed neural network–based anomaly detection systems using TensorFlow and Keras, implementing LSTM and transformer encoder architectures for sequential pattern recognition.
• Built NLU pipelines for customer service chatbots using BERT-based models, applying transfer learning and domain adaptation to improve intent classification accuracy by 20%.
• Designed feature engineering pipelines for large-scale structured and text datasets using dimensionality reduction, embeddings, and representation learning techniques.
• Implemented production ML pipelines with hyperparameter tuning, cross-validation, and statistical significance testing to ensure model robustness.
• Optimized deep learning inference pipelines achieving 40% latency reduction via model compression, TensorRT optimization, and efficient batching.
• Processed large-scale datasets (100M+ records), handling data quality, class imbalance, and train/validation/test splits.
• Mentored junior data scientists on deep learning architectures and production ML best practices.

Data Analytics Specialist - Fidelity Information Services

(2013-01 - 2016-01)

• Developed diagnostic recommendation models using TensorFlow and Keras, implementing feedforward and recurrent neural network architectures for healthcare prediction tasks.
• Built scalable data preprocessing pipelines for healthcare datasets including feature extraction, normalization, and encoding of structured and unstructured data.
• Conducted model evaluation and performance analysis using precision, recall, F1-score, and ROC-AUC metrics across different patient cohorts.
• Integrated trained ML models into production systems via RESTful APIs, implementing secure authentication and monitoring for real-world deployment.
• Collaborated with domain experts to refine model requirements, feature importance, and clinical validation criteria for AI-assisted decision support systems.

Masters - Computer Applications (AI / ML) - Jain University

Bachelors - Commerce (Business) - Manav Bharti University