Hey there — I’m Chandresh Sanwal, data evangelist with roots in the deep unknown.
I didn’t start in a commercial organization. I started in a government research lab, trying to see through the earth.
My early work was in petroleum reservoir modeling — using seismic and electrical signals to map underground rock formations and identify where fossil fuels might be buried kilometers below the surface. You never get to look directly at what you’re trying to find. You work with signals, patterns, and inference. Over time, you develop a feel for what the data is actually telling you, underneath what it appears to be saying.
That’s still how I approach analytics. The numbers on the surface are rarely the whole story.
From the lab to the real world
I left research because I wanted pace. Over the last 24 years, I’ve worked across industries that look nothing like each other — insurance, consumer lending, telecommunications, e-commerce, retail banking, Fintech — across multiple geographies and with teams drawn from more than 60 nationalities.
Analytics at Speed isn’t about rushing — it’s about knowing the terrain well enough to skip the detours. After 24 years of solving variants of the same core problem across six industries, I know where the dead ends are. That means I hit the ground faster, waste less time on what doesn’t work, and get to the useful part sooner.
What I’ve built
Vodafone India / New Zealand — I joined as the first person in the data analytics function. No team, no processes, no playbook. By the time I left, we had a 10-person team and had delivered more than 50 analytics projects: churn models, retention models, cross-sell frameworks, distribution analytics. Recognised with a Strike Force award at Vodafone’s global forum in Switzerland, and subsequently asked to lead the replication of Vodafone India’s analytics model to Vodafone New Zealand.
Telecom data monetisation — I led a project that took mobile network density data and turned it into a commercial product. The first use case: helping taxi fleet operators position their cars where customers would actually be. We proved the concept end to end.
Value-based capex planning — Built a machine learning model to assess site-level utilisation patterns and translate that into a prioritised upgrade roadmap. Analytics sitting at the strategy table, not just serving operations.
Analytics Centre of Excellence — Grew from zero to 25 people, including the complete insourcing of analytical work previously managed by an external vendor.
Banking analytics — Credit models, NPA monitoring, customer segmentation using RFMC frameworks, a cash management system using game theory and ML, and a real-time credit scoring engine for e-commerce that had to respond in milliseconds in a tightly governed production environment.
I’m still hands-on. That’s a choice.
A lot of people at my career stage stop building things and start managing people who build things. I’ve never stopped writing code, building models, or developing products. That’s not a hobby — it’s how I stay sharp and maintain the credibility to walk the talk.
Recent builds: RHYTHM — an ML-powered Smart SIP engine using ARIMA, SARIMA, LSTM, and a Kelly-inspired allocation strategy, outperforming conventional SIPs by 25–30% on backtested data. An automated forex pipeline scraping live rates from 17+ banks every three hours. RAG-based AI pilots using locally hosted LLMs for contract analysis and codebase navigation. Early experiments connecting Agentic AI directly to Power BI via Microsoft’s MCP server.
What I actually believe
Most analytics teams are good at describing the past. The organisations I’ve seen use analytics well treat it as a function that shapes decisions — not one that explains them after the fact.
On AI: I don’t think it threatens serious analytics professionals — I think it exposes the ones whose value was always just execution. What AI cannot do is walk into a bank that has never invested in analytics, understand the political and structural reasons why, earn the trust of a CFO who’s sceptical, and build something that survives after the engagement ends. That knowledge comes from doing it, getting it wrong, and doing it again — across industries, countries, and decades.
The shift isn’t AI replacing analytics leaders. It’s AI raising the floor for everyone, which means the ceiling — judgment, domain expertise, institutional credibility — matters more than ever.
If the work resonates
I write here because most of what gets published about analytics is either too abstract or too promotional. The use cases on this site are from real problems, in real organisations, with the parts that actually made them hard still intact.
If something you’ve read here connects with something you’re working through — let’s think together →