Esha Sharma.
Data Analyst | Data Science & GenAI
I turn data and AI behavior into clear decisions that hold up, from funnel analysis and A/B tests to LLM evaluation.
Available for work.
About me.
I'm Esha Sharma, a Data Analyst with a growing focus on Generative AI, based in India and pursuing my MCA at ABES Engineering College.
My background is in SQL, Python, A/B testing, and cohort and funnel analysis, which I've applied to e-commerce data, product experiments, and the behavior of AI assistants like Claude, Gemini, and ChatGPT.
I build projects independently, scoped like real team deliverables: a clear question, evidence, and a recommendation.
I'm looking for Data Analytics, Data Science, and AI/GenAI analytics internships where analytical depth and clear communication both matter.
AI PM Projects
99K+
Orders Analyzed
AI Products Tested
What I do.
1.
Data & Analytics
SQL funnel and cohort analysis
A/B experiment design and significance testing
Retention and churn analysis
Dashboards and KPI frameworks (Power BI, Tableau)
2.
GenAI & LLM Evaluation
Comparative testing of AI assistants across task types
Failure taxonomies and trust metrics (e.g., Silent Failure Rate)
Prompt engineering and responsible-AI considerations: privacy, failure handling, confidence signaling
3.
Product Analytics & Strategy
PRDs with RICE prioritization and MVP scoping
Product teardowns and competitive benchmarking
Feature recommendations tied to outcome metrics
Stack.
Notion
PRD and documentation
PostgreSQL
Data analysis
Python
Analytics and experimentation
Tableau
Visualization and dashboards
ChatGPT
LLM evaluation
Claude
LLM evaluation
Experience.
Data Analytics & GenAI Projects · Independent
Independent
Jan 2026 - Present
• Analyzed 100K+ e-commerce orders in SQL and Python, mapped a 4-stage funnel, and surfaced ~30–35% month-2 churn in a Tableau cohort dashboard; delivered 3 fixes projecting a 15–20% retention improvement.
• Ran chi-square testing on a simulated 5,000-user dataset (p = 0.000046), identifying a +23.7% Day-7 retention lift; delivered a Power BI dashboard and a ship/iterate memo.
• Evaluated Claude, Gemini, and ChatGPT across 5 task types, built a 5-type failure taxonomy, and delivered a VP-level executive memo recommending confidence signaling as the highest-ROI reliability lever.
• Built a 5-metric AI trust framework (including Silent Failure Rate and Trust Recovery Rate, target 70%+) and a 3-stage Trust Recovery Framework structured for a product team to implement.
• Wrote a 10-section PRD for a Notion AI Workspace Memory Layer (RICE, 6-week MVP, kill condition) and a Perplexity AI teardown identifying 3 retention problems.
What this work demonstrates.
1. Analytical depth
All projects Quote
Every project starts with real data, user research, Reddit threads, actual product usage. Opinions come after evidence, not before.
2. Business impact, not just dashboards
Funnel analysis · A/B test
Findings are tied to a retention rate, a conversion step, or a trust signal, and each ends in a clear recommendation.
AI Trust research · Notion PRD
Failure modes, confidence thresholds, and privacy controls are part of the analysis from the first draft.
4. Independent ownership
