Aman Kumar Sahu

About Aman Kumar Sahu

Aman is a highly skilled Data Scientist with extensive experience in Machine Learning and a strong focus on delivering innovative, data-driven solutions. Known for his ability to implement robust statistical ML models, he has a proven track record of driving meaningful results across a range of industries.With a deep expertise in Natural Language Processing (NLP) and deep learning, Aman demonstrates a consistent commitment to staying ahead in the ever-evolving world of data science. His work in these domains reflects not only technical proficiency but also a forward-thinking approach to solving complex problems with cutting-edge methodologies.Aman’s passion for continuous learning and his adaptability to new technologies make him an invaluable member of any team. He thrives on tackling intricate challenges and transforming raw data into actionable insights, creating solutions that are not only effective but also scalable and impactful.Beyond his technical acumen, Aman is known for his collaborative spirit and ability to inspire those around him. His dedication to delivering high-impact Machine Learning projects is matched by his enthusiasm for mentoring and sharing knowledge, fostering a culture of innovation and excellence.Whether working on predictive modeling, text analytics, or advanced AI applications, Aman’s unwavering dedication and expertise ensure the success of every project he undertakes. His work is a testament to his passion for pushing the boundaries of what’s possible in data science.

Experience

1. Consultant - Data Scientist (Gen Ai)

   KPMG

   Sep 2025 — Present · Maharashtra, IN

   AI-Powered Transaction Monitoring• Developed an AI-driven transaction monitoring pipeline for procurement and Accounts Payable (AP) controls, supporting compliance and spend governance.• Implemented RAG-based structured extraction to standardize data from invoices, purchase orders, contracts, and supplier pricing sources.• Built a multi-stage similarity matching engine (exact, fuzzy, TF-IDF, embeddings) to accurately reconcile invoice line items with contract terms.• Enabled explainable detection of pricing variances, contract non-compliance, and missed discounts, producing audit-ready outputs for AP review. Contract Data Extraction & Intelligence• Architected and deployed a CRAG-driven structured contract extraction system using Azure OpenAI and Azure AI Search to extract 20 compliance-critical attributes from 2,400+ unstructured contracts.• Designed four retrieval indexes (Small, Medium, Large, Extra-Large) to optimize chunk granularity, latency, and cost-performance tradeoffs. • Implemented a hybrid search strategy combining keyword and semantic retrieval, ensuring precise matches for standardized fields while capturing complex, contextually similar clauses. • Developed a 135-record annotated benchmark dataset and applied controlled upsampling to address field sparsity. Maintained prompt and model version control via GitHub to support reproducibility and governance. • Implemented retrieval validation with source-level traceability to ensure auditability and regulatory compliance. • Achieved >90% accuracy and recall across five complex contractual fields, delivering structured, compliance-ready, and audit-traceable outputs aligned with enterprise AI risk standards.

Education

• K. J. Somaiya Institute of Technology

  B.Tech - Bachelor of Technology, Information Technology

• Kendriya Vidyalaya Bhandup,Mumbai

  HSC, Computer Science

• Indian Institute of Technology, Madras

  Advance Certificate in Data Science (BS Degree), Data Science

• Indian Institute of Technology, Madras

  Diploma (BS Degree), Data Science

• Kendriya Vidyalaya Bhandup,Mumbai

  Senior Secondary, PCM

• Indian Institute of Technology, Madras

  Foundation Degree, Data Science