Georgios Sakkas


PhD candidate at UC San Diego
Contact: george.p.sakkas@gmail.com

Update I am on the industry job market! I am looking for research or software engineering positions on program synthesis, program repair, LLM code generation, intelligent coding assistants etc. to start October 2024.

About Me

I am a PhD candidate in the Computer Science & Engineering (CSE) department at UC San Diego, where I conduct research in Programming Languages with my advisor Prof. Ranjit Jhala. My current research focuses on developing tools that make software development, programming, and debugging easier. Specifically, I work on Automated Program Repair and Synthesis, aiming to create fast and reliable tools by combining traditional Programming Languages (PL) research with state-of-the-art Machine Learning (ML) techniques.

Previously, I did my undergraduate studies on Electrical and Computer Engineering at National Technical University of Athens, where I also worked on my thesis on “Resumption Monad Transformers and their Applications in the Semantics of Concurrency” under the supervision of Prof. Nikolaos Papaspyrou.

My name is Georgios or Yiorgos (Greek: Γεώργιος or Γιώργος) but you can also call me George. I grew up in Kalamata, a sleepy beach town in Greece. In my free time, I enjoy playing basketball or going out with friends. I also occasionally do some oil painting and drawing.

Work & Research Experience

A comprehensive list of my work and research experience is available in my most recent resume: Georgios Sakkas’ CV

UC San Diego (U.S.A., Sep. 2018 - Present)

Microsoft Research (Redmond, WA, U.S.A., Jun. 2022 - Sep. 2022)

Amazon.com (San Francisco, CA, U.S.A., Jun. 2021 - Sep. 2021)

Bloomberg L.P. (New York, NY, U.S.A., Jun. 2020 - Aug. 2020)

National Technical University of Athens (Greece, Nov. 2016 - Jul. 2018)

Rutgers, The State University of New Jersey (U.S.A., Jul. 2016 - Aug. 2016)

Publications

My most recent publications include:

1. Title omitted for double-blind review

2. Exploring the Effectiveness of LLM based Test-driven Interactive Code Generation: User Study and Empirical Evaluation

3. Seq2Parse: Neurosymbolic Parse Error Repair

4. Interactive Code Generation via Test-Driven User-Intent Formalization

5. Type Error Feedback via Analytic Program Repair

6. PABLO: Helping Novices Debug Python Code Through Data-Driven Fault Localization

7. InFix: Automatically Repairing Novice Program Inputs

8. AIRMS: A Risk Management Tool using Machine Learning