Photo of Tim Dumol

Hi. I'm Tim Dumol.

I love solving problems using code, math, and algorithms. I am experienced in site reliability engineering, data engineering, algorithms and programming contests, and web development. I have won national and international contests in algorithms & data science. I'm a geek at heart, and love to obsess over my hobbies: baking, board games, electronics, reading, and video games.

Right below are a few highlights of my career. Want to see more? Check my  LinkedIn and my  Github.

Math & Algorithms

  • 174 problems (99.7th percentile worldwide) on Project Euler, a site where you can solve math problems that require programming to solve
  • 636 problems solved (99.8th percentile) on UVa Online Judge, an online algorithmic programming contest practice site
  • Placed in contests, national and international:
    • 1st place, 4g25b Codejam 2013 (Philippines)
    • 1st place, ACM ICPC Philippines March Invitational Contest 2013 (Philippines)
    • 1st place, CompSat Inter-University Programming Competition 2011 (Metro Manila, Philippines)
    • 3rd place, CompSat Inter-University Programming Competition 2011 (Metro Manila, Philippines)
    • 9th place, ACM ICPC Jakarta Regionals 2012 (Southeast and East Asia)
    • 15th place, ACM ICPC Hanoi Regionals 2012 (Southeast and East Asia)
  • Helped found the National Olympiad in Informatics Philippines (NOI.PH) in 2014 and incorporate it in 2017. NOI.PH is the official PH qualifier for the International Olympiad in Informatics, the most prestigious international algorithmic programming contest at the high school level. NOI.PH trains the top competitors in advanced algorithms and data structures, and sends the top delegates to the IOI. Through our efforts, the PH team has received medals each year in the IOI.

Data Engineering

  • Spearheaded Kalibrr's effort to ingest internal data sources into a central BigQuery data warehouse; reducing analysis time in certain use cases >100x from days to minutes
  • Streamlined bringing ML models into production by creating a standard Connexion-based template and Helm chart, improving development velocity 6x from 1 released model per year to 1 per 2 months.
  • Improved process of productionizing ML models by requiring (and creating tooling for) performance benchmarks, Swagger/OpenAPI templates, and automated Cloud Build-based tests.
  • Improved observability of ML models by incorporating error reporting, distributed tracing (via Opencensus/Stackdriver Trace), and Prometheus metrics in the standard Connnexion-based template.
  • Implemented cost control measures for BigQuery usage through a cost dashboard and automated cost alerts on a per-user basis

Site Reliability Engineering

  • Led Kalibrr's move from VPSes + Ansible to Kubernetes, giving us an immutable infrastructure with autoscaling, internal DNS, and blue-green deployment
  • Reduced Kalibrr's GCP bill by 50% in a cost optimization project, mostly through benchmarking and right-sizing requests
  • Enabled observability by setting up alerting & metrics via Prometheus & Grafana, distributed tracing via OpenCensus & Stackdriver Trace, and error reporting; decreasing Mean Time To Repair by making it easier to diagnose incidents rapidly, and making it easy to optimize hotspots by allowing deep dives into endpoint execution.
  • Transitioned container building from Jenkins to Cloud Build, increasing reliability and parallelism for container builds.

Data Science

  • Won 1st place at Trend Micro Codinsanity 2014 (Asia) by using Latent Dirichlet Allocation to perform customer segmentation
  • Won 2nd place at Trend Micro Big Data Innovation 2013 (Asia) with an Apache Storm-based tool to monitor brand sentiment bya analyzing Twitter streams
  • Took online courses: 6.BDx Tackling the Challenges of Big Data (MIT Professional Education); Machine Learning (Coursera)

Web Development

  • Migrated Kalibrr codebase from Javascript to Typescript, fixing and preventing dozens of bugs on the way.
  • Performed profiling and optimization on Kalibrr codebase, resulting in 10–1 000x performance improvements (mainly through query optimization and fixing O(n) query problems)
  • Experienced in performant and secure frontend (Angular, SCSS, Bootstrap, Bulma) and backend (Flask, Pyramid, SQLAlchemy) web development