Daniel Speyer

Research Projects

• Rapid re-identification of human samples using portable DNA sequencing -- Using Nanopore Sequencers to identify individuals from whom DNA was taken. With Sophie Zaaijer, Assaf Gordon, Robert Piccone, Simon Cornelis Groen, and Yaniv Erlich. I devised the core algorithm and did the first test; the coauthors took it from there. Published in eLife. Previously prepublished as Democratizing dna fingerprinting
• DNA.Land -- A framework to collect genomes and phenomes in the era of abundant genetic information. With Jie Yuan, Assaf Gordon, Richard Aufrichtig, Dina Zielinski, Joseph Pickrell and Yaniv Erlich. I contributed various features to the DNA.Land project. Published in Nature Genetics.
• Using mobile sequencers in an academic classroom -- A case study of using nanopore sequencers in class. I was a student in the class, so this is more research on me than by me, but I'm technically on the coauthor list, and my code is offered as an example of student work in such a class. Since I was the only one to solve the second "hackathon", my code probably qualifies as a cherry-picked example, but what can you do? Published in eLife.
• Is there a Firefox Gene? Short answer: no. Long answer: searching for one reveals a lot about what genetic research is like
• CNN Trojan Defense -- Inserting Trojan Horses into Convolutional Neural Nets, and then detecting them.
• Causal Inference -- Inferring Causality from Finite Data using Conditional Independence, with special interest in microbiome disruption and Ileal Crohn's Disease.
• Recursive Systemic Profiler -- Profiling systems which spread across many processes
• SpatialTable -- A Bigtable-style distributed database for spatial information and rectangle queries. Faster than geohashing at the 99th percentile.
• Semi-Byzantine Reputation Networks -- In a world with many bad agents who are willing to lie in a reputation system, can MCMC techniques estimate whom we should trust?. The approach looks promising, but impractical without extensive training data that includes ground truth.
• Alignments -- A way of computing [DNA] sequence alignments more efficiently than the normal dynamic programming approach, for sequences that are pretty similar to begin with. The raw source code is also available.
• ForNet -- This is a distributed logging system for large IP networks. I worked on it along with Professor Nasir Memon, Kulesh Shanmugasundaram, Amos Wetherbee and Radion Khait. Others have worked on the project at other times as well. The project has its own webpage and we wrote a report about it.
• Semi-Automated Refactoring -- This was a project I did as part of a class, and as such is not entirely complete. The idea was to detect sections of code that can and should be combined into single functions to decrease bloat. The software succeeded in finding some such code, but it is not really ready for practical use. You can read my report or you can download the software.

Contact Information

• Email: dspeyer@gmail.com
• GPG Public Key: as text, as binary GPG keyring. (Please do not encrypt email to me by default. To keep my private key safe, I keep it in a place that's somewhat inconvenient for me to get at.)
• Phone: 646-506-5378 (I usually don't answer for unfamiliar numbers, but I do read texts and Signal messages)
• Google Voice: 646-504-5250
• Facebook Messenger: daniel.speyer

Photos

In case you want to know what I look like

Credentials

• Resume [PDF]
• Columbia University Transcript
• University of Maryland Transcript

Rationality

• The Fallacy of Large Numbers -- When handwaving about large numbers just won't do.
• Statistics Calculator -- Doesn't think for you, but does crunch the numbers (and gives you a few reminders of what you should be thinking about).
• Exploring Moralspace -- A first crack at metaethics.
• Against Rationalization II -- Tools to avoid rationalizing
• No Royal Road to Knowledge; No Mercy for Ignorance