Headline: Analytics Strategist & AI Engineer | Ex-Uber, PhonePe, Amazon
Bio:
8.5 years of experience building and scaling complex tech ecosystems from 0-to-1. As a hybrid Data Scientist, Product Analyst, and Data Engineer, I specialize in unlocking hyper-growth and driving operational excellence.
🚀 Why Hire Me Over Other Freelancers?
Massive Scale Experience: I don’t just work on local data models; I’ve optimized systems that serve 60 million+ users and handled high-stakes AI automation for platform integrity.
Execution + Strategy: I’m just as comfortable in the boardroom defining marketing strategy as I am in the IDE building NLP models and data pipelines.
Zero Hand-Holding: Having led analytics initiatives at top-tier firms, I possess the business acumen to jump into your project, identify leaks in your product or data funnel, and fix them autonomously.
What I bring to your project:
Scale: Developer-facing platforms & 60M+ user consumer scaling.
Tech: Advanced AI engineering, NLP, automation, and pipeline construction.
Strategy: Data-driven marketing, product analysis, and 0-1 launch strategy.
If you are looking for a high-level strategist who can also roll up their sleeves and write the code to execute, let’s talk.
Experience Level
Language
Work Experience
Education
Qualifications
Industry Experience
Turning Support Data into Quality Signals: The Launch of BugFinder
I am excited to introduce BugFinder, a transformative, AI-driven platform designed to fundamentally reshape how we surface, cluster, and resolve software defects from our massive global support data.
Currently, identifying engineering defects hidden within customer support channels relies on highly manual parsing. This baseline makes it difficult to uncover “silent” bugs that may be overlooked by support agents or to find complex patterns across disparate, unstructured ticket data.
BugFinder revolutionizes this ecosystem by transitioning Uber from a reactive bug-tracking model to an automated, proactive product intelligence pipeline. By leveraging advanced Large Language Models (LLMs) and multi-layered semantic AI, BugFinder seamlessly synthesizes over ~0.5M daily support tickets into highly actionable, pre-clustered bug candidates. This automation unlocks significant operational leverage, allowing our triage specialists to shift from manual ticket scanning to deep-dive root-cause validation.
Core Strategic Objectives
Drive High-Signal Defect Discovery: Autonomously source at least 10% of all validated engineering bugs directly from raw support interactions via BugFinder’s automated pipeline.
Achieve Intelligent Detection Parity: Match or exceed the defect-detection accuracy of our manual operations, ensuring zero signal loss while radically reducing time-to-surface for critical production anomalies.
Optimize Operational Capabilities: Maximize the productivity of our human-in-the-loop triage teams by automating the repetitive parsing of ticket noise, shifting human capital toward high-value, complex diagnostic validation.
Preserve Engineering Velocity: Ensure downstream Engineering teams receive highly actionable, deduplicated data packages backed by multiple real-world ticket examples, keeping false positives near zero
What Problem We’re Solving
Currently, identifying engineering defects hidden within millions of daily customer support interactions relies on manual, ticket-by-ticket scrubbing. This baseline process creates distinct operational vulnerabilities:
The “Silent Bug” Blind Spot: Support agents frequently resolve technical issues with localized, short-term workarounds. While the immediate customer ticket is closed, the underlying system bug is never escalated to engineering, leaving it to silently impact future users.
Fragmented Macro-Patterns: Manual review processes focus on isolated cases, making it impossible to cross-reference and connect patterns across thousands of disparate complaints. As a result, systemic, large-scale platform issues remain invisible and undetected.
Triage Latency: Traditional manual routing inherently creates a bottleneck. This reactive posture allows emerging software anomalies to compound in production over days, multiplying customer friction before a fix can be prioritized.
Project Overview: InsightAgent
Objective
To automate data-driven insights for the Uber Localization Team by deploying InsightAgent, an LLM-powered virtual data analyst. The agent streamlines monthly performance reporting, ad-hoc data retrieval, and complex root-cause analysis (RCA) regarding translation volumes, workflows, and expenditures.
Approach
I engineered a specialized AI agent directly integrated with our database The architecture relies on three core pillars:
Deterministic Directives: The agent operates strictly on factual database execution—zero assumptions—utilizing optimized, aggregate SQL generation to maintain speed and safety.
Standardized Workflows: Structured templates route user intents into specific paths, whether executing multi-step RCA hypotheses or providing general performance reporting.
Robust Guardrails: Advanced error-handling mechanisms gracefully handle ambiguous inputs, performance bottlenecks, and empty data returns, formatting everything into executive-ready Markdown.
Why It’s Better
Operational Efficiency: Automates manual report generation, reclaiming valuable hours every week for Program Managers and analysts.
Decision Velocity: Provides leadership with instantaneous, consistent, and structured deep-dives to back in-time strategic decisions.
Financial Guardrails: Continuously flags operational anomalies—such as improper project routing or shifts in Cost Per Word (CPW)—safeguarding our bottom line.
Hire a Data Analyst
We have the best data analyst experts on Twine. Hire a data analyst in Bengaluru today.