Paurush Singh

About Paurush Singh

Product & Strategy Executive | Digital Transformation | Financial Services

Contact Information

US Contact

paurush1@outlook.com
650 444 7181
Mountain View, CA
Visa: Australian citizen eligible for E-3 visa

Australian Contact

paurush1@outlook.com
0432 598 857
Sydney, NSW
Status: Australian Citizen
linkedin.com/in/paurushsingh/
Experience Summary

Product management and consulting leader with 18+ years of global experience across 10+ countries in financial services, startups and consulting. I have a proven track record in leading cross-functional teams, building zero-to-one products, and scaling high-impact solutions that serve millions of customers enhancing NPS by over 25%, increasing revenues by more than 20%, and improving profitability by 10% - 50% annually in large banks (balance sheet size of $100bn - $1trn) and medium sized organizations.

Education

Master of Business Administration (MBA)

Indian Institute of Technology Kharagpur, India • 2008

Bachelor of Engineering, Electrical Engineering

Delhi College of Engineering, India • 2004

Key Certifications
Certified Scrum Product Owner (CSPO)
CFA Level 2 Pass
Deep Learning by Andrew Ng
Machine Learning - Stanford University
Digital Product Management - University of Virginia
Core Competencies & Skills

Core Competencies

Product Management
Product Strategy
Business Strategy
Strategic Planning
Customer Experience
AI Product Management

Tech Stack

Python
SQL
Machine Learning (ML)
Tableau
PostgreSQL
QGIS & ArcGIS
Vercel
v0 & Cursor
Figma & Canva
Miro
Replit
JIRA & Confluence
MS Office Suite

Gallup Strengths

Responsible
Collaborative
Analytical
Focus
Adaptable
Key Achievements

Led the launch of a first-of-its-kind geo-spatial product for NAB, driving strategic decision-making and growth enhancing RoE by 2% (from current 12% on a portfolio of $350bn+) over long term.

Founded Oluko, a sports platform, demonstrating entrepreneurial expertise in zero-to-one product development.

Managed large-scale B2C product initiatives for FS clients at Nous Group, improving profitability by 20%+ through innovative product strategies.

Directed global operations at Wipro Digital, delivering a 20% improvement in product delivery performance.

Developed digital loan products at Deloitte, improving NPS by 75% and reducing TAT by 80%.

Selected as one of 150 participants (out of 200K+) for "Genesis Park", PwC's exclusive global leadership development program.

Download Full CV
Get the complete detailed resume with full work experience, projects, and achievements.

Choose the version that best fits your region and requirements.

US Resume

Download US CV

Mountain View, CA • E-3 Visa Eligible

Australian Resume

Download AU CV

Sydney, NSW • Australian Citizen

Insights & Perspectives

Thought leadership on the intersection of AI, finance, and strategic transformation

Deep Dive into LLMs like ChatGPT
Comprehensive exploration of Large Language Model architecture, training methodologies, and real-world applications in modern AI systems.
GRPO Explained: DeepSeekMath
Deep dive into Group Relative Policy Optimization (GRPO) and how DeepSeekMath pushes mathematical reasoning boundaries in open language models.
RAG Survey: Comprehensive Analysis
Detailed examination of Retrieval-Augmented Generation paradigms, covering Naive RAG, Advanced RAG, and Modular RAG approaches for enhanced LLM performance.

This survey offers insights into RAG evolution and implementation strategies for modern AI applications.

Download Paper
SongBloom: Verse by Verse Generation of Full-Length Songs
A novel framework for full-length song generation that leverages an interleaved paradigm of autoregressive sketching and diffusion-based refinement, achieving performance comparable to state-of-the-art commercial platforms.

Key Insights: SongBloom combines high fidelity of diffusion models with scalability of language models for coherent song generation with harmonious instrumental and vocal elements. The framework introduces a verse-by-verse generation approach that maintains musical coherence across full-length compositions.

Download Paper
Training Language Models to Follow Instructions with Human Feedback
Groundbreaking research on aligning language models with user intent through fine-tuning with human feedback, introducing InstructGPT and demonstrating significant improvements in helpfulness, truthfulness, and harmlessness.

Key Insights: Shows how reinforcement learning from human feedback (RLHF) can make language models better at following instructions, with 1.3B parameter InstructGPT preferred over 175B GPT-3 despite being 100x smaller. Demonstrates the effectiveness of human preference optimization for AI alignment.

Download Paper
DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
First-generation reasoning models trained via large-scale reinforcement learning, demonstrating that reasoning capabilities can emerge purely through RL without supervised fine-tuning, achieving performance comparable to OpenAI-o1.

Key Insights: Introduces DeepSeek-R1-Zero and DeepSeek-R1, showcasing remarkable reasoning behaviors like self-verification and reflection that emerge naturally through reinforcement learning processes. Demonstrates breakthrough performance on mathematical reasoning benchmarks.

Download Paper