Shyam Sundaram: Head Digital Innovations & Transformation at Bajaj Finance at Cypher 2017
summary
Overview of practical state of AI: commercial projects run on Julia and notebooks with TensorFlow and H2O plus Python/R libraries and cloud IoT stacks; models split into batch versus always‑learning real‑time systems; vision classifiers often return low confidences (examples 0.6, 0.4, 0.3 or 0.99) and handwriting OCR ranges ~60–90% accuracy; core requirements are repeatability, trustworthiness and bias detection; clear enterprise targets include finance, transportation (early driverless), healthcare and knowledge graphs, with ethics, policy and governance needed before full autonomy.
AI-written summary · about
chapters (10)
- 0:00Start
- 0:15Agenda
- 2:22Privacy and Ethics
- 3:36How Do You Deal with Large Use Cases
- 5:31Julia Workshop
- 11:49Model Layer
- 14:07How Do I Define a Signal
- 23:57Early Stage Discussions on Ai Policy
- 26:17Complex Use Cases
- 26:36Personal Robots
related talks
20mAI : Solving challenges and driving Innovation across Industries
16mOperationalizing AI in Finance: Real-Time Analytics
19mAI: The current state of play
33mAI for Good: Transforming Industries with Ethical & Innovative Solutions
56mData, AI, Hype and Fear - Recipe for Disaster?
44mPanel Discussion: The BFSI Playbook - Scaling AI with Trust and Quality
30mTrustworthy AI - opening up the black box
44mArtificial Intelligence Reloaded - AI Apps in the Industry