Design of algorithms for combinatoric problems; combinatorial chemistry; numeric approximation; convex optimization; non-convex optimization (SGD, Metropolis-Hastings, Gibbs sampling); dynamic programming; branch and bound; transformer-based regression; GAN & WGAN; cache-aware algorithms/bit twiddling

Linux/UNIX command line (e.g., gcc, gdb, valgrind); emacs; Amazon EC2; git; basics of statistical theory; object oriented development; very basic javascript and CSS

Wrote low-latency C++ for use in options trading. Benchmarked techniques at nanosecond level and developed strategies for improving latency of core trading systems. Found and made >10% speedup to overall trade latency, which is applicable to all company trading engines. Scored 92.6% on options trading final exam.

Solved problems and created machine learning tools and models at scale. Programming in C++, Python, basic GPU/multi-GPU computing, using Amazon EC2. Pioneered successful ML strategy for custom Ab design: As first ML hire, designed core ML stack and models (sole author as of Aug 2022); the current version of this ML platform is now branded as "AlphaBind®". Built other critical infrastructure and maths capabilities at a Series A startup company. Designed and implemented the company's generative AI infrastructure for antibodies. Scaled inference to >1B Ab sequences. Designed and implemented new algorithms for solving combinatoric problems that arise in Ab engineering, DNA synthesis, expression, discovery, and statistical analysis. Contributed to a successful fundraising cycle. Discovered simple generalization of Wasserstein distance using higher moments. In passing, invented novel, simple linear algorithm for selection on a list. Designed company t-shirt for 2021.

Wrote object oriented software in C, C++, and python. Developed new algorithms for combinatoric problems, including a novel, optimal algorithm for selection on Cartesian products. My implementation (written with students) of this algorithm powers the current fastest exact isotope calculator. Instructor of record for algorithms, optimization of C++ code, and cybersecurity. Created basic C++ vector class for primitive data that is ~7x faster than std::vector. With student, wrote template-recursive, AVX512 Strassen-Winograd implementation outperforming BLAS. Mentored graduate and undergradaute students. Taught six unique course preps. Wrote >700 pages of publicly available course notes, including on algorithms and on C++. Created The Junior Guide to Combinatorial Chemistry, a client-side interactive webpage that teaches basic programming from up to dynamic programming and teaches basic chemistry and physics up to relativistic mass defect. Awarded over 1.8M USD in total grant funding (direct + indirect).

Taught three unique course preps. Mentored graduate students. Wrote object oriented software in C++. Created C++ cache-oblivious bit-reversal method for template-recursive FFT. Created new moment-based approximation algorithm for building fast semiring algorithms like max-convolution. Created C++ engine for inference on probabilistic cardinal graphical models. Wrote template-recursive C++ library for efficient operations on tensors.

Wrote software in C# and C++. Built dynamic programming algorithms for cardinal graphical models, shipped in Thermo Fisher's Proteome Discoverer™. Optimized existing code. Prepared a patent, and helped protect against infringement of existing IP.

Wrote object oriented software in C++ and python. Created de novo mass spectrometry algorithms. Wrote and implemented a Dirichlet-based goodness of fit measure for determining the location of diminishing returns in false discovery rate curves.

High-performance C and C++ (>10 years); Python (>10 years: package creation and some familiarity with packages numpy, torch, pandas, matplotlib, re, scipy, tqdm, Bio, sklearn, lxml); C# (2 years [industry], rusty); Java (>3 years [student], quite rusty); LaTeX (>10 years); bash (>10 years, basics); Have used but not fluent with: css, javascript/jquery, SQL, and Django.

Studied basic genetics and convex optimization. Created Bayesian approach to mass spectrometry protein inference. Wrote object oriented software in C++ and python. Invented new algorithm for solving LP. Designed an algorithm for genotyping polyploids and used it to make the first profile of the sugarcane genome.

Learned basics of circuit design, relevant mathematics, and algorithms. Created basic circuits. Won freshmen engineering design competition (mini-C compiler). Wrote a neural network package using dynamic programming for efficient backpropagation on deeper networks. Wrote code in Java and C++.