At Tech Mahindra, I engineered the UKG Customer Pain Index, a GenAI analytics pipeline on GCP that analyzes customer, product, account, and support-case data. Gemini 2.5 Flash and Pro power contextual analysis and recommendations, and the solution contributed to a 1% reduction in customer churn over one quarter.
I'm currently productizing the CPI analysis into a stakeholder-facing analytics product, bringing pain scores, themes, trends, and recommended actions into one interface. I also automated Google Maps Street View geospatial workflows, reducing Africa-wide coverage analysis from one month to one day and supporting a USD 2M+ business opportunity.
I developed an internal domain-specific LLM fine-tuning POC using Qwen2.5-1.5B, QLoRA, SFT, and DPO, comparing model outputs for correctness, relevance, and hallucination behavior. At HSBC, I automated product-risk scoring and monitoring and developed a backtesting methodology for the Standard Initial Margin Model.
At Credit Suisse, I built Python and SQL automation for price testing, IPV, reconciliation, and controls, and developed market-data extraction APIs. My work spans GenAI, machine learning, customer analytics, risk modelling, and data automation.

