CV

Contact

• jack [at] jbkjr.[first 3 letters of “commercial”]
• john.koch [at] yale.[first 3 letters of “education”]

Education

• Yale University, New Haven, Connecticut, USA
  • B.S. in Applied Mathematics
    • Thesis: Tree-to-Tree Semantic Parsing
      • Adviser: Prof. Dragomir Radev
    • Concentration: Computer Science.
    • Awards: George J. Schulz Summer Fellowship in the Physical Sciences
    • Relevant Coursework: Artificial Intelligence, Data Mining & Machine Learning, Natural Language Processing, Deep Learning Theory & Applications, Advanced Natural Language Processing, Selected Topics in Neural Nets, Data Analysis, Optimization Techniques
• Stanford University, Stanford, California, USA
  • Summer Program
    • Classwork in Stochastic Processes and Real Analysis

Related Experience

• Language, Information, and Learning at Yale, Yale University
  • Undergraduate Research Student, Nov. 2017 to May 2018, Sep. 2018 to Dec. 2018
    • Developed and implemented machine learning models in the domain of natural language processing
    • Utilized tree-based architectures for semantic parsing (Spring 2018)
    • Developed and tested a new “Temporal Capsule Net” architecture on Visual-QA tasks (Spring 2018)
    • Selected to supervise undergraduate research applying the Transformer architecture to text-to-SQL tasks (Fall 2018)
    • Applied and compared unsupervised pre-training techniques to text generation task of semantic parsing (Fall 2018)
  • Research Assistant, Jun. 2018 to Aug. 2018
    • Applied the Transformer architecture to language modeling tasks
    • Aided initial data scraping and analysis for new “BioNLP” project in conjunction with Yale School of Medicine

Projects

• Comparing Pre-trained Language Models with Semantic Parsing
  • First application of pre-trained language models to semantic parsing and the first application of the recent ELMo, OpenAI GPT, and BERT language models to any language generation task
  • Most significant project undertaken individually
• Tree2Tree: A Reimplementation and Discussion of Tree-Based Neural Semantic Parsing
  • Compared sequence-to-sequence, sequence-to-tree, and novel tree-to-tree architectures for semantic parsing
  • Contributions to this project satisfied senior thesis requirements

Skills

• Computer Languages: Python, PyTorch, TensorFlow, C, Java, LaTeX, R, Racket

Interests

• Consumer technology, cinema, philosophy.

References

• Professor Dragomir Radev, Yale University
  • dragomir.radev [at] yale.[first 3 letters of “education”]