Dippu Kumar Singh is a senior director of emerging data and analytics at Fujitsu North America, developing production AI systems for demanding enterprise environments. His work spans contact-center voice intelligence, computer vision, healthcare decision support, and the governance required when sensitive data drives consequential decisions.
Before focusing on generative AI, Singh designed large-scale data platforms and streaming pipelines using Spark, Kafka, and Hadoop, and developed data architectures for manufacturing and public-sector organizations. His work also includes markerless motion analytics, protecting computer-vision systems against data poisoning, and contributions to Fujitsu’s AI ethics and compliance assessments.
Turning difficult signals into dependable decisions
- VoiceOps for contact centers. Singh’s voice-intelligence architecture converts overlapping customer-service conversations into structured records of customer intent, agent actions, and resolutions. Multichannel capture, transcription, language-model extraction, and CRM synchronization target after-call work reduction while giving managers more consistent operational data.
- Privacy before model processing. Separate customer and agent audio channels preserve speaker attribution; noise filtering, specialized vocabulary, and numerical normalization improve transcription. Early PII masking prevents sensitive details from reaching downstream models unnecessarily, although additional safeguards introduce latency and architectural complexity.
- Structured outputs with human oversight. Few-shot prompts, predefined intent categories, transcript-grounding checks, and JSON schemas make generated information usable by business systems. Human-in-the-loop verification keeps operators responsible for correcting and approving records. Singh’s proposed extensions include explainable coaching, demand-based staffing forecasts, and early detection of abusive calls.
