Designing and shipping AI agents, analytical pipelines, and automation tools that multiply team impact.
View ProjectsEnd-to-end case analytics pipeline turning raw Salesforce support data into structured monthly insights. Built a 9+ category symptom taxonomy, automated Claude & GPT-4.1-mini classification pipeline, and interactive HTML dashboard for the LinkedIn Measurement product area — reducing report production from 4–6 hours to 30 minutes and eliminating tribal knowledge loss at role transitions.
Used LinkedIn's internal AI analytics agent (engined by Claude) to build a dual-view interactive OKR dashboard — a manager-facing status view and a personal individual contributor competency-tagged view — sourced live from Google Sheets and Docs. Produced promotion-ready evidence packages through natural language interaction.
Led bug bash and cross-functional coordination for LinkedIn's AI-powered Campaigns' Audience feature during a leadership gap. Identified critical bugs affecting Strategic clients, opened Jira tickets, and created the Accelerate Audiences one-sheeter — ensuring Go-to-Market continuity.
Presented at MCXO Hackathon 2025. Proposed a Glean-based agent to identify duplicate PORT Jira tickets before creation, saving estimated 20 min/instance across ITS. R&D expressed interest in productionalizing the solution — opportunity sizing engagement initiated with LinkedIn Business Platform POM team.
Glean Workflows agent automating discovery, filtering, and packaging of sales-relevant product updates for the biweekly ITS newsletter. Searches defined sources with sales-facing keywords, categorizes into "Need to Know," "Coming Up," and "Tips," and outputs consistent summaries. Positive C2 pilot feedback ("Wow it's even grouped by the categories of the newsletter. This is so clever").
AI agent that inspects healthcare advertiser domains (homepage, About, Products/Services, professional sections), applies LMS Help Center and internal policy rules, and outputs consistent B2B vs B2C decisions with embedded next steps (Legal escalation or rejection). Strong SME endorsement ("this is so cool").
Agent that reads Slack escalation threads and outputs a JIRA-ready package for our team to send to Product Technical and Engineering teams: a structured Jira ticket (title, description, expected vs actual behavior, references). Eliminates manual copy-paste handoffs and enforces consistent documentation quality.
Takes user-listed objectives, pulls org-chart role context and project artifacts, writes SMART key results, buckets each by impact/probability (Big Bet / Home Run / Small Win / Junk), and outputs a structured planning table — replacing hours of manual review and formatting.
Calendar-aware peer feedback agent that determines the correct semester/annual review window, pulls collaboration artifacts from the period, and produces structured compassionate feedback (start/stop/continue) aligned to LinkedIn's review framework with supporting resource links.
Glean AI agent that automates average-clicks-per-day calculation across campaigns, normalizes engagement over time, and outputs structured performance summaries with multi-campaign table comparison — eliminating repetitive manual calculations in support workflows.