Last year, , where I worked along side the Customer Operations and Analytics team (now known as the Innovation and Sustainability team) to solve operational inefficiencies within our workflows, softwares, and departments.
Highlights
- My largest project at BC Hydro focused on reducing the average handle time within the call centre. Using Excel, Power BI, Power Query, and DAX Studio, I recovered and processed 50% more of our lost call centre data by joining and cleaning 600K+ records, then delivered a proposal around agent training, performance tracking, and system gap identification. Visualizing the results showed a path to lift agent efficiency by 5% and save $50,000+ a year through reduced customer handling time, at zero additional cost once fully integrated.
- My second project focused on the Interactive Voice Response (IVR) system. Analyzing the same dataset, I found a misrouting issue that sent 50% of callers to the general queue after failed verification, where repeated transfers and re-verification by multiple agents added 5 minutes to each interaction. By analyzing Google Dialogflow call data for high-frequency intent phrases, I improved utterance recognition and reduced the error rate from 30% to 25% across 4,000 calls, projected to save 150,000+ minutes a year at scale.
- Nearing the end of my time at BC Hydro, I helped build a financial forecasting model that incorporated various growth assumptions, cost drivers, and scenarios to project changes in Canada Post expenses under different rate and fee structures. With Canada Post expenses exceeding $10 million YTD, the model supported financial planning, reporting, and executive level decision making.
- One of the most impactful experiences for developing my software and analytical skills was supporting BC Hydro's Call for Power Energy Management Portfolio initiative, which enabled owners of multiple properties to consolidate their accounts and access preferred energy rates. To improve application quality and reporting accuracy, I reconciled data for 1,500+ external companies across two unstructured Excel trackers and submission databases, creating a reliable dataset for categorization, trend analysis, forecasting, and management reporting.
- The most enjoyable experience I had at BC Hydro was organizing a team building event as an intern. At the time, three separate teams had recently been merged into one, creating some ambiguity around roles and responsibilities. I led team exercises to identify which responsibilities belonged to each department, encouraged open communication among team members, and helped coordinate a team outing afterward. As part of the event, we participated in a guided history tour of Vancouver's Chinatown led by Judy Maxwell, where we learned about the history and cultural significance of the neighborhood.
Skills & Tools
- Excel
- Power BI
- Power Query
- DAX Studio
- SAP BI
- Google Dialogflow
- Forecasting
Chinatown, and the calls that went nowhere
My time at BC Hydro really came down to two things I keep coming back to. One was about people, and the other was about a system that quietly wasn't working. They happened months apart, but I think about them together, because both were really the same problem: something was getting lost between the people involved.
The people one first. Three separate teams had recently been merged into a single group, and nobody was quite sure where one team's responsibilities ended and the next began. As the intern, I organized a team-building day to work through it. I ran exercises to map which responsibilities actually belonged to which department, pushed for open conversation about the overlaps people had been quietly working around, and then coordinated an outing afterwards so the day didn't end on a whiteboard.
We spent that afternoon on a guided history tour of Vancouver's Chinatown led by Judy Maxwell, learning about the history and cultural significance of the neighbourhood. It was easily the most enjoyable thing I did there, and it did more for how the team worked together than any of the exercises I'd planned.
The system one was the Interactive Voice Response line. Half of our callers were failing verification and landing in the general queue, where repeated transfers and re-verification by multiple agents added 5 minutes to each interaction. I worked on utterance recognition in Google Dialogflow and brought the error rate down from 30% to 25% across 4,000 calls, which projected to save 150,000+ minutes a year at scale that were being spent sending people to the wrong place.
What stuck with me is that the second problem was invisible until someone went looking for it. Nothing was on fire. The line worked, calls got answered, and a quarter of the people on them were still being misrouted.