2 Years Postdoctoral or Engineer Position

2 year Postdoc/Eng position

2 Year Postdoctoral or Engineer Position:
Differential Privacy in PostgreSQL

INSA Centre Val de Loire is offering an opportunity for either a Postdoctoral Researcher or Engineer
diverse profiles and backgrounds, as we believe that both types of expertise can bring valuable contribu
- tions to our project. The specific responsibilities, objectives, and scope of the position will be tailored to
to start as soon as possible, but we are flexible with the start date to accommodate the right individual’s
availability.

Host. Petscraft project-team, with a flexible location either at INSA Centre Val de Loire, campus
Bourges or Inria Saclay centre, Turing building in Palaiseau.

Application. Please contact us if you are interested or need additional details. Application can be sent

Context. This position is part of the project ”Making PostgreSQL Deferentially Private for Transparent
AI” funded by the french national research agency ANR. The project involves several French research
teams working on data protection, from Inria, universities and engineering schools, as well as Dalibo,
company expert in PostgreSQL and developing PostgreSQL Anonymizer [1, 2, 3]. The general objective
of the project is to propose, implement within PostgreSQL and evaluate a privacy preserving approach
for interpreting SQL queries in the sense of differential privacy (DP). These queries will range from the
Select-Project-Join-Aggregation (SPJA) form to the export of releases (DUMP) of a part of the database
in order to be able to work on it as if it contained no sensitive data. Specifically, the main objective is
to extend the anonymization models already integrated in PostgreSQL Anonymizer (pseudonymization,
k-anonymization and addition of noise) to other models verifying DP, as well as DP existing and novel
relaxations.

Objectives and roadmap.
Anonymizer with the most promising (in terms of utility) approaches [4, 5, 6, 7] to answering SPJA
queries that guarantee DP (or similar models, such as local differential privacy). The main lock of this
task is to design this DP mechanism as an extension of the SQL grammar in order to qualify a specific
query or function with DP metadata.
wide range between implementation and theory. They may include more theoretical contributions, e.g.
regarding novel DP relaxations or the impact of semantics and schema on DP. In terms of implementation,
they could include, for example, the development of demonstrators, validation and evaluation platforms
for the novel models proposed in the project.

Computer Science. The following skills are appreciated:

- Knowledge in privacy & anonimization.
- Knowledge & experience in databases and database management systems, particularly SQL and

PostgreSQL.
- Proficiency in programming.

Dalibo will provide training on PostgreSQL and PostgreSQL anonymizer. Courses on anonymization
and differential privacy can be proposed at INSA CVL.

Salary & benefits. The salary range is 34000 to 38500 €/year, depending on experience, skills, and
qualifications. Benefits:

- Subsidized meals
- Free public transport in Bourges
- Possibility of teleworking (2 days/week)
- Professional equipment available (videoconferencing, loan of computer equipment, etc.)
- Access to vocational training
- Social security coverage
- 7 weeks of paid leave per year

2 year Postdoc/Eng position

References

[2] Clochard, D.: Postgresql anonymizer, canal-u. In: Contraintes réglementaires liées aux bases de

io/postgresql_anonymizer/how-to.handout.pdf

[4] Dong, W., Fang, J., Yi, K., Tao, Y., Machanavajjhala, A.: R2t: Instance-optimal truncation for
differentially private query evaluation with foreign keys. In: Proceedings of the 2022 International
Conference on Management of Data. p. 759-772. SIGMOD ’22, Association for Computing Machinery,
3514221.3517844

[5] Johnson, N.M., Near, J.P., Song, D.X.: Practical differential privacy for SQL queries using elastic

[6] Kotsogiannis, I., Tao, Y., He, X., Fanaeepour, M., Machanavajjhala, A., Hay, M., Miklau, G.: Pri
- vatesql: a differentially private sql query engine. Proc. VLDB Endow. 12(11), 1371-1384 (Jul 2019).

[7] Wilson, R.J., Zhang, C.Y., Lam, W., Desfontaines, D., Simmons-Marengo, D., Gipson, B.: Dif
//arxiv.org/abs/1909.01917