Missy
Most of my industry work is closed-source, so I describe those systems only as far as the work allows. For names and dates, see my resume.
Publications
- A Review of Topic Modeling: From Probabilistic Foundations to Neural and Multimodal Approaches
Esparza, J. (2026). Preprint, Zenodo. doi:10.5281/zenodo.20280741
- Understanding the Issues Surrounding COVID-19 Vaccine Roll Out via User Tweets
Esparza, J., Bejarano, G., Ramesh, A., & Seetharam, A. (2021). Computational Data and Social Networks (CSoNet 2021), Lecture Notes in Computer Science, pp. 197–205. Springer. doi:10.1007/978-3-030-91434-9_18
- Convolutional Neural Networks in Respiratory Conditions Detection: A Systematic Review of the Last 4 Years
Esparza, J. (2020). Research poster, Summer School of Machine Learning (SMILES 2020), Skolkovo Institute of Science and Technology.
Production systems
I build a LangGraph agent system on FastAPI and Next.js: a ReAct supervisor over a self-critiquing four-sub-agent reasoning graph, plus ten specialized agents that integrate GitHub, Confluence, Monday.com, Slack and Google Drive. For retrieval I combined vector and full-text search with RRF/MMR reranking, chunk-level access control and provider failover. I also wrote the Redis job queue and WebSocket notifications behind the interaction-capture engine that feeds the CRM.
For customer support on WhatsApp I built a LangGraph agent grounded in a Postgres/pgvector knowledge base, AWS-backed OTP delivery, audio transcription with faster-whisper, and monthly financial reports that an LLM drafts on a schedule. I added an MCP server that exposes internal tools to LLM agents, and BI dashboards modeling interchange, GMV, spreads and user growth.
I built a transformer recommender that surfaces semantically adjacent lessons at sub-second latency, so a query for "tomato" returns grocery shopping and meal preparation next to the direct matches. I also fine-tuned Llama 3.1 8B with LoRA for curriculum-resource generation, grounded in internal technical documentation; expert reviewers confirmed it held the curriculum's voice.
I led five engineers on AWS ETL (Glue, Athena, Lambda, S3) that ingested product-availability data from the B2B platforms of Cencosud, Falabella, Oxxo, Soriana and Walmart, and fed the dashboards our clients used for their decisions.
Open-source software
NVFP4 (W4A4) quantization of LatamGPT's 70B model on a single 32 GB workstation.
A privacy-first voice companion that runs a small LLM on a laptop CPU, with no GPU: faster-whisper, Qwen3.5-4B on llama.cpp, Piper TTS, Chroma retrieval and a persona LoRA. I built it for the Hugging Face Build Small hackathon.
A machine learning framework for Julia.
Experiment tracking for machine learning in Julia.
Pure-Julia gradient boosting with native categorical feature handling and symmetric decision trees.
Google Summer of Code 2023. I brought MLflow experiment tracking to the Julia ecosystem.
A Julia client for the MLflow REST API.
Auto-rotating perceptrons in Julia.
A CSS framework in the Pluto.jl aesthetic.
A PSP HUD plugin in C.
Painless proton-ge installation for Steam Deck and Linux.
Further repositories are on my GitHub profile.
Notebooks
I reimplemented classic teaching notebooks in Julia with Pluto.jl, which keeps them reactive and reproducible.
MLOps & Lifecycle Management
High-Performance Computing & GPU Programming
Deep Learning Architectures
Data Engineering & Scalable Processing
Honors and awards
Skolkovo Institute of Science and Technology · August 2020
Skoltech picked me from more than 2,000 applicants to present a systematic review of convolutional neural networks for respiratory-condition detection.
Research Experience for Peruvian Undergraduates · October 2020
I joined as an undergraduate participant and worked under Gisella Bejarano in Arti Ramesh's research lab at Binghamton University.
Organization of American States · May 2019
The OAS gave me a place in Creando una Trayectoria Profesional en Seguridad Digital, which included training at Universidad Federico Villarreal in Lima.
Academic service
Research Experience for Peruvian Undergraduates · April 2026
I evaluated undergraduate research presentations across scientific disciplines and wrote feedback for each participant.
IEEE · July 2024
I reviewed paper submissions for technical accuracy, originality and fit with the conference program.