Paurush Singh

Paurush Singh

Product leader and strategy executive in financial services

Summary

Senior financial services leader in product management and consulting with 15+ years of experience building and scaling digital products across banks, insurers and fintech. Proven track record of translating strategy into execution, launching zero-to-one products and delivering measurable improvements in customer experience, revenue and portfolio growth.

Focus areas

Lending and credit

Loan origination, credit decisioning, risk-based pricing and portfolio growth

Portfolio strategy

Geospatial risk, climate exposure, concentration and opportunity mapping

AI and data products

Scenario modelling, eligibility engines and decision-support tooling

Digital channel execution

Conversion optimisation, digital journeys and application funnel design

Zero-to-one product builds

Founding, scoping and launching net-new product lines end to end

Financial services transformation

Operating model change, cross-functional delivery and capability uplift

Selected achievements

Led the launch of a first-of-its-kind geospatial product for NAB, enhancing RoE by 2% over long term on a portfolio of $350bn+.

Founded Oluko, a sports platform, demonstrating zero-to-one product development and entrepreneurial execution.

Managed large-scale B2C product initiatives for financial services clients at Nous Group, improving profitability by 20%+.

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 time-to-approval by 80%.

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

Experience summary

Product Manager | Independent Consultant / NAB (National Australia Bank)

NAB · 2023 – Present

Geospatial portfolio intelligence, lending products, digital transformation

Director, Strategy & Product

Nous Group · 2022 – 2023

Financial services strategy, customer experience, B2C product management

Director, Design-led Transformation

Wipro Digital · 2021 – 2022

Product delivery, cross-border teams, fintech and banking clients

Founder & CEO

Oluko (Sports Platform) · 2019 – 2021

Zero-to-one product, fundraising, team building

Director, Strategy and Operations

Deloitte · 2017 – 2019

Loan origination, digital journeys, NPS improvement

Associate Director, Strategy and Operations

PwC · 2008 – 2017

Strategy consulting, Genesis Park leadership program

Education

Master of Business Administration (MBA)

IIT Kharagpur, India · 2008

Bachelor of Engineering, Electrical

Delhi College of Engineering, India · 2004

Certifications
Certified Scrum Product Owner (CSPO)
CFA Level 2 Pass
Deep Learning — Andrew Ng
Machine Learning — Stanford University
Digital Product Management — University of Virginia
Get in touch
linkedin.com/in/paurushsingh/paurush1@outlook.com

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