Tristan Chong

EDUCATION

Classification algorithms, language modeling, part-of-speech tagging, parsing, tokenization, named entity recognition, regular expressions, word sense disambiguation, information retrieval, formal grammars, automatic summarization, dialogue systems, speech recognition & synthesis

Phonetics, phonology, morphology, syntax, semantics, pragmatics

WORK

I spent the latter half of my employment with PagerDuty in the Product Development org, contributing both as Full-Stack Developer and Data/ML Engineer on the Event Intelligence team.

• Led the architectural planning and development of Similar Incidents, a feature that provides relevant historical context to incident responders, from personal hackday project to general availability
• Built, productionized, and operated the machine learning models, data stores/pipelines, backend services/APIs, and frontend web user interfaces that comprise PagerDuty's various Alert Grouping offerings, reducing noise and repeated interruptions for related issues during triage and remediation
• Provided expertise in chat command interface design and development while on loan to the Platform team, delivering a first-class integration with Atlassian HipChat (RIP) and laying the groundwork for future integrations
• Coached and fostered the growth of interns into full-fledged contributors within an agile development setting

In my first 2 years at PagerDuty, my time was split evenly between 2 roles in our Infrastructure org: first, in a capacity I would characterize as a Release Engineer on the Developer Tools team, and then as a Site Reliability Engineer on the eponymous team.

• Facilitated the company's transition from a monolithic architecture to microservices by building chat-based tooling around deployment, configuration management, and multi-cloud provisioning
• Drove the organization-wide adoption of modern, immutable infrastructure by educating and supporting teams through the process of containerizing their services
• Enabled the product and engineering organizations to iterate more quickly through the build-measure-learn feedback loop by introducing continuous integration and delivery
• Automated the creation of new production-ready skeleton services compliant with best practices around monitoring, logging, and secrets management, resulting in improved developer happiness and faster time to market
• Built a simplified abstraction layer to allow teams to run applications in a highly available and scalable manner across multiple environments, integrating open-source service discovery, load balancing, and container orchestration technologies into an internal platform
• Led the SRE team's response to various major incidents during regular on-call shifts, including the NTP case study featured in Google's SRE Workbook

• Accepted a short-term contract role to update the data pipeline and NLP-related services I had built with my former team to run on the company's new containerized infrastructure

• Implemented signup and authentication via Facebook
• Built a database of potentially interested customers by extracting information from user Facebook profiles and mobile phone contacts (if shared) on signup
• Enabled the marketing team to filter prospective users by various criteria with a custom application backed by the aforementioned database
• Automated the sending of email invitations to targeted prospects
• Developed tools to simplify common tasks with Amazon Web Services APIs

• Built a pipeline utilizing the Stanford CoreNLP software suite to parse tens of millions of pages of text
• Extended the capabilities of the parsing pipeline to scale up and down automatically by provisioning and decommissioning AWS EC2 instances in accordance with load
• Implemented a service-oriented architecture designed to extract and cache data on named entities, syntactic heads, coreference chains, dependency relations, and sentiment
• Wrote ETL and load balancing modules in a Python library used for data science research
• Researched various document summarization algorithms, and evaluated n-gram keyword extraction & sentiment analysis for potential business applications
• Developed a heuristic to infer the subject of a wiki using term frequency and weighted scoring
• Trained latent Dirichlet allocation (LDA) models with named entity data and used a distance metric to identify related pages as part of a recommendation system

• Worked as part of a team of linguists and engineers to develop machine translation software and spoken dialogue systems
• Wrote context-free grammars in a Backus-Naur Form variant for parsing natural language
• Tested and tuned support vector machines (SVMs) and semantic class taggers to achieve higher phrase classification accuracy
• Created training corpora using Python and the Natural Language Toolkit (NLTK) to automate tasks including crowdsourced data collection, text normalization, production of canonical forms via stemming and lemmatization, morphosyntactic operations, and elicitation of use cases for regression testing
• Developed YAML interaction guides to manage dialogue states and conversation flow
• Localized existing applications to different languages, regions, and target markets
• Optimized synthesized speech audio using noise removal, silence trimming, normalization and compression techniques
• Translated text-to-speech pronunciation dictionaries between X-SAMPA and proprietary formats