Headline

Business Analyst II @ Meesho | Data Engineering & Analytics | Spark · Databricks · Airflow · SQL · Python | Built AI-powered root-cause analysis used by 100+ analysts

About

I build data systems that explain themselves. At Meesho, I sit in the CXO Office's Commerce Platform Control Tower, owning DAU and NMV diagnostics for a 100M+ user marketplace. My job is to answer the question leadership asks every morning: what moved, and why? Instead of answering it by hand, I've been automating it: - An AI-powered RCA engine on Databricks that decomposes metric movements (LMDI, MECE L1–L4 attribution, 24-check validation gate). In a blind evaluation it scored 4.1/5 against 3.3 for analyst-written RCAs, and it's now used by 100+ people across 3 charters. - A daily anomaly-alerting system that catches DAU and order-rate deviations, runs the diagnosis, and posts an owner-assigned ticket to leadership before anyone asks. - A growth analytics app on Databricks Apps with plan-vs-actual tracking, dimension cuts, waterfalls, and one-click RCA. Before that, I covered search ranking, catalog quality and user growth analytics at Meesho. At PharmEasy, I built production ETL pipelines with Python and Apache Airflow. What I work with: Python · SQL · PySpark · Databricks · Apache Airflow · ETL/ELT · Data Modeling · LLM applications · Superset · BI & automated reporting · GitHub · CI/CD What I'm good at: sitting at the boundary between engineering and analytics. I can build the pipeline, make it reliable at scale, and explain what the numbers are saying to people who'll act on them.

Location

Bengaluru

AI stack

claude code

cursor

hugging face

notebooklm

openai

huggingface

Work experience

Business Analyst II at Meesho, Dec 2025 - Present · 11 mos, Own DAU and NMV diagnostics for a 100M+ user marketplace in the CXO Office – Commerce Platform Control Tower. • Built and shipped a Databricks Genie RCA skill (LMDI decomposition, MECE L1–L4 attribution, 24-check validation gate), rolled out to ~70 Growth/NMV/MAU users; in a blind eval it scored 4.1/5 vs 3.3 for human-written RCAs. • Designed an automated D-1 anomaly alerting system that detects DAU/order-rate deviations, runs the RCA and posts findings with an owner-assigned ticket to leadership's Slack channel daily. • Built a growth analytics app on Databricks Apps (KPI tiles, plan vs actual, dimension cuts, waterfall charts, one-click RCA deep-links). • Built a supplier-ticket classifier on 4.3 lakh tickets (TF-IDF + SGD, 76.9% accuracy) with Spark UDF inference; mapped 81+ raw dispositions into 34 themes.

Business Analyst I at Meesho, Jul 2024 - Dec 2025 · 1 yr 6 mos

Data Science & Analytics Intern at PharmEasy, Jan 2024 - Jul 2024 · 7 mos, Bengaluru, Karnataka, India, • Built Python-based ETL pipelines to automate ingestion and transformation of operational datasets • Used Apache Airflow to orchestrate and schedule data workflows for analytics use cases • Optimized complex SQL queries, improving reporting performance and query efficiency • Implemented GitHub-based version control and pipeline management for data workflows • Generated analytical insights supporting product and operational decision making

Research Intern at Indian Institute of Information Technology Vadodara, May 2023 - Jul 2023 · 3 mos, Gandhinagar, Gujarat, India

Google DSC Cloud at Google Developer Student Clubs IIIT Vadodara, Dec 2021 - May 2023 · 1 yr 6 mos

Education

Bachelor of Technology - BTech, Computer Science, Indian Institute of Information Technology Vadodara

Science, Fergusson college

Public links

github: https://github.com/2pushkaraj3

codeforces: https://codeforces.com/profile/Pushkaraj_

github: https://github.com/2pushkaraj3