Jesse
Chumo

Selected work

01

Wera

2026 / Go

A self-hosted job radar. It watches the job boards of 570+ companies, filters new postings with rules, and has an LLM score each one against a resume.

wera-frontend-five.vercel.apphover to scroll

The interesting problem is cost. Postings are fetched once for everyone, unchanged boards cost nothing, and most jobs are ranked or ruled out locally before a model gets involved, so the LLM reads each posting once rather than once per user.

1. 01

   Fetch boards

   ETags on every request. An unchanged board answers 304.

   0 bytes if unchanged
2. 02

   Store changes

   Content hash per posting; only new or changed rows are written.

   shared by all users
3. 03

   Rule filter

   Role, seniority, location and sponsorship rules, per user.

   40k jobs in 4 s
4. 04

   Local ranking

   TF-IDF against the profile, so a new user sees ranked jobs in seconds.

   no model
5. 05

   Shared facts

   One LLM call per posting version: seniority, years, sponsorship.

   1 call, every user
6. 06

   Fact exclusions

   Too senior, too many years, no sponsorship: dropped before scoring.

   about half drop out
7. 07

   Fit score

   A compact job card and profile, with a cached prompt prefix.

   ≈ $0.0004 per job
8. →

   Ranked list

   Today, jobs, industries and an application tracker.

New user sees ranked jobs

~5 s

was 5 min

Cost of a fit call

$0.00038

was $0.00075

Re-filter 43,732 jobs

4.1 s

was 9.3 s

Ruled out before any fit call

22 of 46

sample run

Measured on development data.

Do GNN-based vulnerability detectors survive obfuscation?

Thesis Evaluating the Robustness of GNN Vulnerability Detectors Under Code Obfuscation

int sum(int n) {    int total = 0;    for (int i = 0; i < n; i++)        total += i;    return total;}

int f_3a(int v_1) {    int v_9 = 0, v_2 = 0;    if (v_1 < 0 && v_1 > 0)        v_9 = v_1 * 7;    while (1) {        if (v_2 >= v_1) break;        v_9 += v_2;        v_2++;    }    return v_9;}

GraphSentinel is the service I built to flag these transformations automatically, so experiments can run over whole datasets.

Experience

1. Aug 2025 — present

   Systems Administrator

   University of Texas at Arlington Arlington, TX

   • Managed VirtualBox lab environments on RHEL hosts supporting Windows, Fedora, and Kali Linux VMs and their snapshots for hands-on cybersecurity instruction, enabling reliable and repeatable student security experiments.
   • Owned and maintained Linux-based lab infrastructure and two department RHEL production servers supporting 1,000+ users, diagnosing virtualization and software issues to ensure stable service delivery for students, faculty, and researchers.
   • Designed and deployed an AI experimentation lab using RAG, LangChain, and LangGraph, enabling students and researchers to run LLM experiments on applied AI workflows.
   • Built Python tooling to automate software provisioning and installation workflows across lab machines, eliminating repetitive manual setup and improving consistency across environments.

   RHEL / Linux / VirtualBox / Python / RAG / LangChain / LangGraph

2. Jan 2025 — Aug 2025

   Graduate Teaching Assistant

   University of Texas at Arlington Arlington, TX

   • Achieved a 15% improvement in average student performance for the Algorithms course by providing tutoring and office hours focused on core concepts.
   • Reduced recurring coding errors and improved code quality by 20% by delivering detailed feedback on C programming assignments and reinforcing best practices.
   • Elevated student satisfaction to 90% by facilitating collaborative debugging sessions and structured code reviews.

   Algorithms / C / Code review

3. Aug 2023 — Aug 2024

   Full Stack Developer

   Fasi Health Remote

   • Built and maintained full-stack features across patient, driver, and admin workflows using NestJS, TypeScript, React, GraphQL, and PostgreSQL, supporting core business operations in a production healthcare platform.
   • Developed a React Native/Expo mobile app for a ride-hailing service, translating UI designs into functional screens and integrating REST and GraphQL APIs for real-time data flow.
   • Engineered backend infrastructure including NestJS services, GraphQL resolvers, Prisma migrations on a Dockerized PostgreSQL database, AWS Cognito for authentication, S3 for document and media storage, and payment platform integrations.
   • Built CI/CD pipelines with GitHub Actions and AWS Elastic Beanstalk, reducing deployment time by 50% and improving release reliability.
   • Contributed to Agile delivery through sprint planning, code reviews, and merge request approvals, supporting consistent and reliable feature releases.

   NestJS / TypeScript / React / React Native / GraphQL / PostgreSQL / Prisma / Docker / AWS / GitHub Actions

4. Nov 2022 — Jul 2023

   Front End Web Developer

   Fameve Remote

   • Reduced production issues by 60% by implementing automated testing and improving code quality in an Angular-based e-commerce platform.
   • Developed modular React and Angular components for core e-commerce workflows, consuming REST APIs for data rendering and collaborating with backend engineers to ensure reliable end-to-end feature delivery.

   Angular / React / REST / Automated testing

5. May 2022 — Nov 2022

   Software Development Intern

   JamboPay Nairobi, Kenya

   • Built REST API endpoints for a fintech payment platform using NestJS, supporting transaction processing workflows in a production environment.
   • Designed and optimized SQL queries to surface transaction data for admin reporting, improving operational visibility into payment workflows.

   NestJS / REST / SQL

About

                       :::::::::
                   ::.......:::::...
                  .........,::;--===-;.
                :..,;-=+=+++++===++++**=.
               :..:;:::;----====++====+**.
               ...;::;;-=++++#%##@@@@@%##*,
              :..,:;;-=++*###%@@%@@@@@@@%%*
              :.,::;:;=++*#%%%%%%@@@@@@@%#%:
               ..:;::;=*##%%%@@@%@@@@@#***#:
               ..:-;:;=;,...,;+**#+;,.;+###,
               ..;=-:..;.:#...;+@@*=:.. :+@#-
             ,,+.;-==;.;,-@+;=*#@@@@@+*%@@@#-
             .,+.;;=#@@#***@@**#@@@@@@@@@@@%
              -,:,;-+%@@@@@@==*@@@@@@@@@@@@=
               +*.,;-+#@@@#,,;:,;%*%@%*#@@@;
                ;.,;;;;-+**#%*-=*+@@@@@#+#*
                 ;.:::::-+*+++**#%%#*+@@@%
                  :.,::;-;,.,:=+%%@@%*#@%;
                ,.,,..,,::,:::;*@#%##@@#
               .  .-;,...,;-;,..:=+%@@*
             ;... :++-;.....-+%%#*#+-+.,
            ;...,..:=+=-::..........=@..+
        ::..........,;-----;:,.,+@@@@@..+=-
    ::::,,,,,........,:;=*##++*#@@@@%..++==++--
 ;:;;;;;----;::,........,-=*%%#%@@@+,;*=-===++==+=-
;;;---=-:;--=;;-;;;,...... ..+@#*-;,..+=;-======+=+==-;
;;;--;-=-,,;-=-----=-. ,,....,%;..-. .-=;:======+-;-===+-:
::;-;-;--::;:;;;;;;-===;-;----:.+++=-:;-=-======+;;==--===
,:,;-;;;=-;::;;--;;;;-==-;;;;=;,*==--+-=========-;--=;----
,;..:=-::;--;;--==-;;;;;-=----.;*=-;-============--------:
,;,..:=-;;;--=======----;--==-.-;=--===========---------:,
:,,...;-;;;;;;;;-=========--=-.-:;=---========----------,;
::....,::,....,..:--;--===--=+.;*==-=----====----------;;;
,,....,,...,:;;;;,:-;:;------=,,+=---------------------;;;
,.. .,,,.,,:::;;;-;;-;;------=:.=--;------------;------;;: