Jesse Thomas Palma Email & Phone Number

Somerville, Ma, Us

We are currently developing a robust platform for AI-powered heavy machinery maintenance and parts service enablement. We are Digital Iron. Reach out if interested!

San Francisco, Ca, Us

I introduced OpenAI models into the EcoCart core calculations.Single-handedly improved performance of the core AI system from 20% to 75% accuracy in a couple months, and set up as 6ms response API to host.In addition, I collaborated environmental modelers to develop the long-term vision of the product.I also built a data labeling app from scratch, specific to the domain and data.And I supplemented OpenAI models to build a unique system specialized to identify products as a hierarchical taxonomy, and to evaluate them with hierarchical metrics.

Cambridge, Us

I worked on machine learning models that use ocean data to predict long-range weather more accurately than any others in this space.Predictions of percipitation, temperature, hurricanes, etc months in advance. I worked on both improving accuracy and quantifying uncertainty in predictions.

Dallas, Texas, Us

I worked on spatial-temporal models of movement of assets in the supply chain to measure uncertainties, predict arrivals and detect anomalies.

I applied NLP methods to procurement transaction documents as part of a policy-impact model to discover opportunities for companies to save financially and reduce environmentally.

Liverpool Street, England, Gb

I built machine learning methods to analyze maritime and weather data. This included prediction, statistics, classification and modeling of commodity flows. I adopted an innovative approach, applying NLP techniques to maritime data, that produced shockingly effective results.I developed and wrote a provisional patent for the invented method.

Cambridge, Massachusetts, Us

I led research and development of methods for optimization, predictive analytics, and machine learning for smarter and more effective logistics. I used a variety of techniques including Big data (Spark), Bayesian statistics, Markov probabilities and Neural networks. When a single data source is weak, we bootstrapped diverse data sets with each other to find better solutions for tracking and predictions to support logistics decisions. I also developed aggregate indices to provide clients with an understanding of performance of ports and facilities.

Boston, Ma, Us

I was a research scientist in the Neuromorphics Lab. I acted as the primary computational modeller and PI for an Air Force Research Lab (AFRL) funded project, in collaboration with Neurala, Inc. The project, titled "Adaptive Laminar Computing on Scalable Heterogeneous Hardware for Image Understanding" lead to the conception of a novel symbiotic fast and slow learning system. The conceived learning design has been submitted for provisional patent by Neurala. It combines Adaptive Resonance Theory whole-pattern fast learning with slower part-based learning for compressed sparse coding.For over a year, I led a weekly journal club on Deep Learning methods for the lab.

Boston, Ma, Us

The focus of my effort as a researcher was to advance the understanding of the learning mechanisms of memory and stability in brain circuits. Culminating in my doctoral thesis, my work under Stephen Grossberg examined the stability of short-term memory of patterns in cortical and subcortical circuits under neuromodulation. (described in more detail under education)During this time, I also acted as a Teaching Assistant and presented lectures for several graduate-level courses.

Mclean, Va, Us

I worked at the MITRE Corporation for 3 years as a software engineer, as well 3 summers as an intern, and part-time during graduate school for a year. My projects spanned a variety of government applications. This included work on Air Force target tracking algorithms to software for intelligence analysis. I worked on projects aimed to improve data visualization and foster collaboration. I expanded my knowledge here on database management with SQL, adaptable UIs, sorting and re-organizing data to accelerate and ease the high information throughput for analysts.

I worked as part of the Mining the Bibliome project - a natural language processing (NLP) project. My primary role was data production, but I also participated in research meetings and wrote a revision of the lab's annotation manual.

I constructed a predictive modelling system in Pascal for a National Imagery and Mapping Agency (NIMA) project. The main product was a tool for analysts to build fuzzy logic networks that could then be automatically optimized by Genetic Algorithm of the logical parameters.

Doctor Of Philosophy (Phd), Neural Networks

Bachelor Of Arts (B.A.), Cognitive Science

High School Diploma

Education record