Track record
Enterprise work: Akamai and Robert Half
Six years as the person between a finance or marketing org and its systems: requirements, configuration, UAT, and rollout, usually across many markets at once. Titles, dates, and numbers are exact.
Akamai TechnologiesBusiness Systems Analyst · Cambridge, MA · Sep 2019 – Dec 2022
Functional owner of Oracle EBS R12 Procure-to-Pay (PO, AP, Payables, eBTax, iProcurement), with working exposure to GL and AR. About 30 concurrent projects a year, 120+ across the tenure. I started as a co-op in January 2019 and was converted to full time.
Payments: $1B+ a year, automated. Accounts payable was hand-triggering vendor payments across 30 markets. I led the automation independently, no PM: 3 XML host-to-host formats (ACH, wire, FX), BR100 specs and schemas mapped to banking-partner requirements, and an exception-escalation design for negative acknowledgments. It rolled out through 16 SDLC cycles, 9 with JP Morgan and 7 with Citi. 20,000+ invoices a year now flow with success rates above 99.9%.
Tax: 30 markets on one automated platform. Oracle's rules-based tax module needed manual reconfiguration every time a rule changed. I owned the BSA workstream on a 20+ person program, learned indirect-tax determination well enough to define system logic independently, and configured Thomson Reuters OneSource single-handedly. 7 SDLC phases across market groups. $800K in annual savings, $4M over 5 years on the NPV analysis.
Expenses: 8,000+ employees, 99% of compliance checks automated. Replaced Oracle iExpense with SAP Concur across 3 phased rollouts (US, India, then Poland and the remaining markets), with AppZen's AI engine tiering expense reports by risk: low and medium auto-approved and paid, high routed to human review. I worked directly with internal audit to keep every workflow SOX-compliant. To avoid a reporting blackout at cutover, I also migrated 20+ operational reports to IBM Cognos on the same go-live timeline, learning Cognos from scratch to do it.
Also designed the pandemic-era home-office stipend flow with Oracle as source of truth and Concur as source of record, which avoided custom code entirely.
Robert HalfManager, Systems Analysis · Boston, MA · Dec 2022 – Mar 2025
Hired to build the systems-analysis function for a 40-person global marketing analytics org spanning 5 verticals. 1 direct report, 2 project managers dotted-line.
Call intelligence: 65% to 90% accuracy. 680,000+ calls a year came in with no reliable way to tell client leads from candidate and admin calls, so ad spend couldn't be attributed to the calls it produced. Invoca was already in place; my job was to make its classification trustworthy. I set a recall baseline by listening to calls, built a confusion-matrix framework, and worked with Invoca's data science team through iterative model refinements: 65% classification accuracy to 90%. A monthly drift-testing program validated every new model before rollout. The eval discipline in the MapChat process → started here.
Operating model: intake from 2 weeks to 2-3 days. Requests reached a 40-person function through email and hallway conversations, and leadership meetings were spent on triage. I built the intake and operating model: every request captured and auto-created in Jira, a DACI accountability framework, and formal handovers between verticals. Approval cycles went from 2 weeks to 2-3 business days, and the leadership meeting got its agenda back.
Governance: one tracking standard for global marketing. Naming conventions, taxonomy, and UTM standards unified across global marketing teams, enforced through utm.io, with training and weekly office hours for 200+ stakeholders. Data consistency improved about 30%, and attribution across $20M in annual spend became reliable enough to act on.
Also ran a solo Confluence SSO migration: 200+ users, 100GB+ of data, zero access disruption, finished ahead of schedule.
Before Akamai
Before the enterprise years: 2 years as a Technical Account Manager at DK Electricals in India, running the full sales cycle for cloud and on-prem systems, about $2M a year. I started on the customer side and worked my way toward the build from there.
The years before and after these: About →