Dhananjay Kumar

About Dhananjay Kumar

I’m a Principal Data Scientist and Senior Manager specializing in enterprise-grade GenAI systems for finance, accounting, and regulated data environments.I help organizations move beyond GenAI proofs-of-concept to production-ready, compliant AI systems—where accuracy, auditability, and trust matter as much as model performance.My work sits at the intersection of:Artificial Intelligence × Accounting × Evidence-Based DecisioningWhat I do best• AI for Financial Reporting & AssuranceDesign and deploy LLM-driven systems for SEC/EDGAR analysis, IPO readiness, private-equity diligence, and accounting workflows, with emphasis on traceability, source attribution, and reviewer-ready outputs.• Enterprise GenAI ArchitectureLead end-to-end delivery of secure, scalable LLM platforms (RAG, fine-tuning, orchestration) integrated with financial data platforms and ERP-adjacent workflows.• Explainable & Defensible AI SystemsArchitect AI solutions with clear data lineage, human-review checkpoints, and governance controls, enabling use in compliance-critical and audit-sensitive environments—not black-box automation.• Identity & Relationship Substantiation for Audit and ForensicsDesign graph-based models to reconcile entities, counterparties, and digital identities across fragmented financial records—supporting related-party identification, control analysis, and evidentiary completeness in audit and forensic investigations.• LLM Engineering & ScaleDrive multi-GPU LLM training and optimization (PyTorch DDP, DeepSpeed), cloud deployment (Azure), and enterprise data integration (Snowflake, Docker, PromptFlow).• Leadership & DeliveryLead cross-functional data science and engineering teams, translating complex AI capabilities into decision-grade outputs with measurable business ROI.

Experience

1. Senior Manager Principal Data Scientist

   Connor Group

   Jun 2020 — Present · Lehi, UT, US

   Production‐Grade AI Product: Delivered a domain‐specific LLM application for the accounting industry, which accelerated compliance checks and streamline financial reporting, cutting manual overhead significantly.• Fine‐Tuning Techniques: Leveraged Parameter‐Efficient Fine‐Tuning (PEFT), Sparse Fine‐Tuning (SFT), and Data Point Optimization (DPO) to tailor models lile Gemma, Qwen, Llama and Mistral for domain‐specific financial tasks. • Multi‐GPU HPC & LLM: Implemented distributed training (PyTorch DDP) using Lambda Labs cloud GPUs to fine‐tune open‐source LLMs (Llama, Mistral). Accelerated training cycles by ~40% and improved domain accuracy for financial tasks.• RAG & Automated Workflows: Built Retrieval‐Augmented Generation solutions for SEC/EDGAR document analysis, boosting IPO readiness and automating accounting intelligence processes.• LLMOps & MLOps: Deployed advanced ML pipelines via Azure ML, Docker (Azure Kubernetes service) and PromptFlow, ensuring smooth version control and continuous model retraining for finance clients.• Predictive Analytics: Led the design of exit‐strategy models (IPO, LBO, acquisition) using data from PitchBook, Crunchbase, and Snowflake, delivering data‐driven insights for strategic business decisions.• Leadership & Collaboration: Managed a cross‐functional team of data scientists and engineers, fostering best practices in HPC/LLM while aligning AI initiatives with financial stakeholders.

Education

• Maharshi Dayanand University

  BE, Electronics & Communicaton

• CUNY School of Professional Studies

  Master of Science - MS, Data Science

Skills

• Web Services
• Sap Brfplus
• Statistics
• Computer Vision
• Jquery UI
• Big Data
• Ic Webclient
• Opencv
• Sap Fiori
• Data Visualization
• Soap
• Sap Netweaver Gateway
• R
• Amazon Web Services (Aws)
• Abap
• Xamarin
• Crm
• Sap Erp
• Marketing Analytics
• Swift
• Badis
• Predictive Modeling
• Ajax
• Android Development
• Tableau
• Talend Open Studio
• Statistical Modeling
• Sap Crm
• Bol
• Sup
• Weka
• Artificial Intelligence
• Cloud Foundry
• Sap Ui5
• Jquery
• Sapui5
• Machine Learning
• Scikit-Learn
• Digital Marketing
• Wsdl