Some info about me.

I am a constant learner.

I came to Machine Learning after getting a PhD in Biochemistry in a Computational Biophysics lab. Conducting biophysical simulations entails learning about optimization techniques, such as gradient descent or simulated annealling, and how to setup their parameters. Sampling techniques such as Markov Chain Monte Carlo are also used in non-dynamic simulations to obtain equilibrium distributions. So there is a lot of overlap between ML and my doctoral training. I found that this training, along with the maths and statistics I learned while getting my BA in Mathematics, smoothed out the ML learning curve.

I care about data quality because data-driven decisions can be wrong.

My scientific training has fine-tuned my ‘data sentience’: using data as input entails a minimum knowledge about its origin (population, method of acquisition, schema), as well as accountability for its pre-processing and usage. This information should be part of a project’s documentation.

I like to show data.

I aim to produce information-rich data visualizations: ones with meaningful scales, markers, accessible color schemes, and labeling. I believe that a visualization should not need decyphering or raise questions because of its design: instantaneous data comprehension is the ultimate ‘prettiness’ criterion for a data picture.

I like to mentor.

I have several motivations for mentoring/taching:

  • I want to be the mentor I never had
  • I’m most satisfied when I help someone become more efficient

I consider teaching a personal growth skill that:

  • Helps me zoom in on the relevant points needed for a particular student to learn about about a particular topic
  • Sets the high-bar of knowledge transfer: if you don’t get it, by dedault the root cause is my explanation

My rants:

  1. Web designers who select absurdly low text or color contrast.
    I consider this a technical bottleneck in the information flow because it is so antithetical to providing information. This also apply to the choice of hyper tiny fonts, which I call ‘legalese font size’.

  2. Unusable public data: this happens when one has to contact the posting agency/organization in order to obtain the field definitions and, possibly, the data structure. This is a case of false transparency usually mendated by a well-meaning law - e.g. “public data should be public”, but lacking a definition for useability.

Foreign languages

  • English
  • French
  • Some Spanish


Graduate research:

  • Article:

    Two Cl Ions and a Glu Compete for a Helix Cage in the CLC Proton/Cl- Antiporter Chenal C and Gunner MR; Biophysical Journal 113, 1025–1036, September 5, 2017 doi: 10.1016/j.bpj.2017.07.025 pdf

  • Thesis:

    Chloride and Proton Binding in the E. coli 2Cl¯:1H+ CLC Exchanger, 2017 https://academicworks.cuny.edu/gc_etds/1874/

Undergraduate research:

  • Article:

    Determination of the Degree of Charge-Transfer Contributions to Surface-EnhancedRaman Spectroscopy Chenal C, Birke RL and Lombardi JR; ChemPhysChem 2008, 9,1617–1623 doi: 10.1002/cphc.200800221

  • Article:

    DFT, SERS, and Single-Molecule SERS of Crystal Violet Cañamares MV, Chenal C, Birke RL, and Lombardi JR; J.Phys.Chem.C 2008, 112(51),20295–20300 doi: 10.1021/jp807807j