Kuruva Radhakrishna

Business Analyst 1

Bengaluru, India 0+ yrs exp 57 · Fair

About

No bio added yet

Skills & Expertise

No skills listed yet

Work Experience

Business Analyst 1 (BA-1)

Meesho

Jul 2026 - Present

Production ran at 6.5–7% delivery breach on sale and disruption days – the single-quantile approach couldn’t cut breach without inflating promised dates, hurting customer experience. Diagnosed the driver mix across pin-lanes, quantiles, sale windows, weather-callout days, and backlog bands to isolate which signals earned a lift and which added noise. Built a per-lane calibration layer on the delivery-date model: fits each thick pin-lane’s confidence level from history, with a network-wide fallback for sparse lanes and added protection for chronically-breaching destinations. Layered contextual lifts for weather severity, high order-backlog days, sale tail-days, and weekends – plus a manual day-buffer ops can inject on specific disruptive lanes. Cut breach from 6.5–7% to < 5% on sale and disruption days while matching production’s promised-date speed; backtested at scale and live at 10% traffic.

Business Analyst Intern

Meesho

Jan 2026 - Jun 2026

Built metric views and an evaluation set for Databricks Genie agents, and contributed to MIA (an in-house replica of Genie Agents) – cut ad-hoc SQL queries by 25%. Automated week-on-week metric reporting via Kaizen on top of Gold-layer data models, eliminating manual metric updates and saving 2–3 hours of team time every week; v1 is live with the team. Authored an RCA playbook for Seller NPS – mapped the analysis flow and the metrics that reveal seller intent, turning ad-hoc investigations into a repeatable framework the team now runs off.

SWE Co-Op Intern

AlgoUniversity

May 2025 - Jul 2025

Built a production-level online judge platform with real-time submissions and multi-language support – Java/Spring Boot REST APIs powering the backend, React frontend, MongoDB persistence. Designed a Docker-isolated compiler service for C, C++, Java, Python – sandboxed per submission, with request queuing and real-time result streaming to the frontend. Integrated Gemini-powered AI for code review, debugging, and conversational assistance during problem-solving sessions. Deployed on AWS EC2 with a custom domain; load-tested the compiler service with k6 at 100 concurrent users / 46 req/sec, achieving zero failures and sub-250ms latency.